Why a Decision Record Matters
An adaptive platform makes decisions that can affect multiple arms, sponsors, sites, or participant groups. A later reader may need to establish not only what was decided, but what question the authority considered, which evidence was available at the time, who had authority to act, and whether the action depended on conditions. Without that context, a correct decision can be difficult to implement consistently, and a decision made outside the proper authority can be mistaken for an approved one.
A decision record is a controlled account of a consequential choice and its traceable basis. It complements, rather than replaces, the procedures for approvals, change control, participant transitions, or sponsor accountability. It also does not turn a recommendation into an approval: the record must identify which body made the decision and which body, if any, only advised it.
The record should let a reader reconstruct the decision as it stood at the time—not rewrite its rationale using later knowledge. It therefore identifies the evidence set and relevant document versions, notes material uncertainty or disagreement, and distinguishes facts from interpretation. The detail needed depends on the decision’s consequences and the platform’s governing procedures, but the basic traceability requirement remains the same.
A Practical Structure for the Record
A consistent structure reduces omissions and makes records easier to locate. A record need not be lengthy if the issue is narrow; it does need to answer each relevant question precisely. Use a stable identifier, date, version, and links or references to supporting materials so that the record can be found and related to subsequent actions.
| Record element | What it should establish |
|---|---|
| Identity and scope | Record identifier, date, platform or arm, decision-maker, and the boundaries of the question. |
| Question | The choice put to the authority, stated neutrally and specifically. |
| Evidence considered | Relevant data summaries, operational information, advice, document versions, and uncertainties available at the time. |
| Authority and process | The body or role empowered to decide, its applicable scope, and the recorded outcome of the required process. |
| Decision and rationale | The action chosen and the reasons linking the evidence and governing considerations to that action. |
| Conditions and limits | Prerequisites, scope restrictions, effective date, review triggers, or matters expressly left undecided. |
| Follow-up and closure | Action owners, deliverables, due dates, completion evidence, and how the record will be updated or linked to implementation. |
The question should not smuggle the preferred answer into its wording. “Should the platform close Arm B because it is ineffective?” presumes both the interpretation and the conclusion. “Given the specified interim evidence and the governing decision rule, should enrollment to Arm B continue, pause, or stop?” identifies the choice without asserting what the evidence proves. The record can reference an already established decision rule rather than restating or altering it.
Evidence should be identified specifically enough to distinguish it from other versions or later analyses. “Interim results reviewed” is weak if the platform has several cuts. A stronger reference states the data-cut date, analysis or report identifier, relevant version, and any important limitation. The record should not reproduce confidential material unnecessarily; controlled references or appropriately restricted attachments can preserve traceability while limiting access.
Worked Example: Recording a Decision About an Arm
Hypothetical situation. On 18 April, a platform committee considers whether enrollment to Arm C should continue. The committee has seven voting members. Six attend; its governing procedure requires at least five members for a quorum. Five members vote to continue enrollment and one votes to pause it. The decision rule and authority for this question are already specified in the platform’s governance documents.
Question and evidence. The record states: “Should enrollment to Arm C continue under the current platform rule?” It identifies the interim report as version 2.1, based on a data cut dated 31 March, and links the committee briefing. It notes that outcome data are still immature and that the report does not address a separate operational concern raised by sites.
Authority and outcome. The record names the committee as the decision-making authority for this question and points to the applicable governance provision. It records six of seven voting members present, meeting the five-member quorum requirement (\(6 \geq 5\)), and the vote as five to continue and one to pause. The outcome is continuation, not a unanimous recommendation. The dissenting vote is recorded without implying that the member’s reasoning was known if it was not provided.
Rationale, conditions, and follow-up. The rationale states that the committee applied the current rule to the identified report and chose continuation, while recognizing the immaturity of the outcome data. It does not claim that continuation proves Arm C is effective or that the operational concern has been resolved. The record specifies any condition actually imposed, such as a scheduled reconsideration at the next review, and assigns an owner and due date for the site concern. If no condition was imposed, it says so rather than leaving the point ambiguous.
This entry preserves the evidence version, authority, process, result, and limits. A later reviewer can distinguish the committee’s decision from the unresolved site concern and can check whether follow-up occurred.
Why the Elements Belong Together
The structure can be justified by asking whether a later reader could reliably answer six questions: What was decided? On what question? Using which information? By whom? For what reason? With what consequences or next steps? If an element is missing, the chain from issue to implementation may break. For example, a vote without a named authority does not establish that the voters had power to decide; a rationale without a dated evidence reference may not establish what information was available; and an action without an owner or due date may remain an intention rather than a trackable commitment.
This is a practical completeness argument, not a claim that a well-filled form guarantees a sound decision. The record makes the reasoning inspectable and actions traceable; it cannot repair inadequate evidence, authority outside its remit, or a flawed decision process. The merits of the choice and the quality of its documentation are related but distinct questions.
State the choice, its scope, and what is not being decided. This prevents readers from extending the conclusion beyond the matter considered.
Reference dated materials and versions, then name the authorized decision-maker and the applicable remit. Record advice separately from the decision.
State the action and explain how the identified evidence and governing considerations support it, including uncertainty or material dissent.
List conditions, owners, deadlines, and the evidence that will show completion. Link later amendments or implementation records without silently replacing the original decision.
Worked Example: Eligibility Change With a Condition
Hypothetical situation. A platform authority considers a proposal to broaden an eligibility criterion for a new cohort. The proposal cites a feasibility summary covering 42 screened candidates at four sites: 11 did not meet the current criterion, and the summary reports that 8 of those 11 might have been eligible under the proposed wording. The authority approves the change only after specified protocol and site materials are updated and the required local approvals are in place.
Evidence and interpretation. The record identifies the feasibility summary’s date and version and reports the counts accurately: \(8/11 = 0.727\), or about 72.7% of the candidates who failed that criterion in this small summary. It describes this as a feasibility signal, not as proof that the change will improve recruitment or preserve treatment-effect validity. It also records that the summary covers four sites and is not presented as evidence from all platform sites.
Authority, decision, and condition. The record names the approving authority and the governance provision establishing its remit. It states that the eligibility change is approved subject to completion of the specified document updates and required local approvals; it does not imply that participants may be enrolled under the revised criterion before those prerequisites are met. The rationale connects the limited feasibility evidence to the decision while acknowledging its scope.
Follow-up. The record assigns the document update to the responsible owner, the site communication to a named role, and a due date for each. It identifies the evidence needed to close the condition, such as approved document versions and confirmation of required site readiness. If a prerequisite is delayed, the condition remains open; the approval is not inaccurately recorded as immediate implementation.
Versioning, Confidentiality, and the Record’s Limits
Records should be dated and version-controlled. If a correction is needed, preserve the original entry or its history and explain what changed, when, and by whom. A later decision should be recorded as a new decision or clearly linked update; it should not overwrite the earlier rationale as though the later information had been available at the time. Referencing controlled protocol versions and change-control records helps keep the decision record connected to implementation without duplicating those documents.
A record may contain sensitive operational, participant, sponsor, or interim information. Access should follow the applicable confidentiality and information-governance arrangements. A restricted attachment, access-controlled repository, or redacted circulation copy can be appropriate, provided the authoritative record retains a clear reference to the evidence and the permitted access route. Redaction must not make it impossible for authorized reviewers to determine what evidence informed the decision.
A decision record is not a substitute for formal minutes where those are required, an approval signature or other required authorization, a protocol amendment, or a change log. Minutes may capture discussion broadly; a decision record isolates and organizes the traceable basis and follow-up for a consequential choice. An email may communicate an outcome quickly, but it is not a durable record unless it is incorporated into the controlled recordkeeping process. Likewise, an evidence reference does not replace the source report.
Worked Example: Documenting a Shared-Data Access Decision
Hypothetical situation. Two sponsors request access to a platform summary containing pooled operational information. The authorized governance body permits release of a specified aggregate table to both sponsors, but not participant-level records or sponsor-specific interim results. The body requires a disclosure check before release.
Question and evidence. The record states precisely which summary and data cut were considered and whether the request was for aggregate or individual-level information. It lists the applicable data-access and confidentiality provisions and notes any uncertainty about whether a table cell could indirectly identify a small group.
Decision and rationale. It names the authorized body, records its outcome, and explains that the approved scope is limited to the identified aggregate table because the request does not justify broader access under the applicable arrangements. It distinguishes the body’s decision from any staff recommendation. It does not state that all aggregate data are automatically non-identifying.
Conditions and follow-up. The record assigns the disclosure check to an owner, states that release must wait until the check is complete, and identifies who will document its completion. It records the released version and date once that condition is met. This creates a traceable path from request to authorized scope to actual release without exposing restricted details in the general record.
Common Mistakes and a Quality Check
- Recording only the outcome. “Arm C continues” does not identify the question, authority, or evidence. Add the decision context and traceable basis.
- Confusing recommendation with decision. A committee or expert may advise, while another authority decides. Label each role and record the actual authorized outcome.
- Writing a retrospective rationale. Do not use later results to make an earlier choice appear better supported than it was. Date the evidence set and preserve the original reasoning.
- Treating a condition as a note. “Update sites” is not sufficiently actionable. Name the owner, deliverable, deadline, and what will demonstrate completion.
- Overstating what evidence shows. A small feasibility summary may inform operational planning, but it does not by itself establish a clinical benefit or validate a changed population.
- Assuming a complete record proves a correct decision. Documentation enables review; it does not establish that the authority, evidence, or judgment was adequate.
Before finalizing a record, check whether a person who did not attend can identify the precise question, the evidence version, the responsible authority, the result and rationale, any limits or conditions, and the owner and timing of follow-up. If any answer depends on private recollection, clarify the record or link the relevant controlled source.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A record says, “The panel supported the proposal.” What key information is needed before this can be treated as a traceable decision record?
- A dated report summarizes 12 cases, and 9 meet a feasibility target. Calculate the observed proportion and state one limitation on its interpretation.
- A decision is approved subject to required site readiness. What should the record say about implementation and follow-up?
- Why should a later decision not silently replace the original record after new evidence becomes available?
Solutions. (1) Specify the question and scope, identify the panel’s authority and whether it decided or only advised, reference the evidence and versions considered, state the actual outcome and rationale, and document conditions and follow-up. (2) The observed proportion is \(9/12 = 0.75\), or 75%. It is a summary of 12 cases, so it should not be treated as a precise estimate for a broader population without considering how those cases were selected and how representative they are. (3) State that implementation must wait until the readiness condition is met; assign an owner and due date, specify what evidence closes the condition, and preserve the status until completion is documented. (4) The original record must show what was known and decided at that time. Record the later decision as a linked update so the chronology and basis of both decisions remain inspectable.