Why Diagnose the Path, Not Just the Duration?
A platform decision that takes six weeks may reflect a stalled process, but duration alone does not explain why it took six weeks. The decision may have required substantive review by several appropriate groups, waited in a queue because reviewers were overloaded, or moved repeatedly between roles because no one had clear authority to resolve it. These conditions call for different remedies. Adding staff will not fix an unresolved question about who can decide; naming an owner will not make an overloaded review group available sooner.
Governance diagnosis therefore begins by reconstructing the path of a particular decision. Decision records and traceable governance provide an important starting point: they can show when a decision was submitted, referred, discussed, and resolved. The diagnostic task is to interpret that sequence without treating a long interval, a referral, or an undocumented step as conclusive evidence on its own.
The focus here is the governance process: who had responsibility, who could make or escalate the decision, and whether the process had sufficient time and people to perform its assigned work. The merits of the clinical or statistical decision are separate questions. A sound diagnosis also distinguishes a genuine bottleneck from a deliberate, documented pause for necessary review.
Define the Decision and Its Timeline
Before measuring elapsed time, define the decision and its endpoints. For example, the start might be the date a complete amendment package was submitted to the designated governance process; the endpoint might be the date the authorized body recorded its decision. If the start instead is the date a sponsor first raised the issue informally, the resulting duration is not directly comparable. Use the same definition when comparing decisions.
A useful timeline records dated events and intervals, not merely a final number. It distinguishes time spent in active review from time waiting for a meeting, a response, missing material, or assignment to a decision-maker. Where the records permit, calculate:
Here, \(T_{\text{total}}\) is elapsed time from the defined start to endpoint. The two components must cover that interval without double-counting: an interval cannot be counted as both active review and waiting time. If activities overlap or the records cannot establish what was happening, report that limitation rather than forcing a precise decomposition. Calendar days and business days are both possible units, but the unit and any excluded periods must be stated.
The decomposition is descriptive, not causal. A large waiting component identifies where time accumulated; it does not by itself establish why the work waited. A meeting may be delayed by limited scheduling capacity, by uncertainty about which body should meet, or by an appropriate requirement for additional information.
Worked Example: Locating Delay Within an Amendment Timeline
Hypothetical data. A platform begins tracking a complete proposed amendment on 1 March and records the authorized decision on 28 March. For this example, elapsed time is counted in calendar days as the difference between dates, so 1 March to 28 March is 27 days. The timeline shows 6 days of active review, then 14 days waiting for the scheduled governance meeting, then 7 days of active review and decision preparation.
The total is \(6 + 14 + 7 = 27\) days. Active review accounts for \(6 + 7 = 13\) days, and waiting accounts for 14 days. Thus the waiting share is \(14/27 \approx 0.5185\), or about 51.9% of the elapsed time.
The meeting queue is the largest identified interval and is a reasonable place to investigate further. It is not yet evidence of inadequate resources: the meeting may have been scheduled at the next available date under an appropriate cadence, or the delay may reflect a rule requiring a particular forum. The next diagnostic step is to establish why the meeting could not occur sooner and whether the assigned decision route was clear.
Read Referrals and Ownership Together
A referral is not inherently a failure. A complex decision may properly require input from a sponsor, an ethics committee, a regulator, a safety committee, or another governance body. The diagnostic signal is a pattern in which the same unresolved question repeatedly moves between roles without a clear account of who will integrate the input and bring the question to resolution.
For a consistent review, count a repeated referral when a decision or a material part of it is sent back to a role or body that has already considered it, because the question remains unresolved. Record the reason and the next action if available. Do not count each routine consultation as a loop: a planned sequence of distinct reviews is different from repeated circulation of the same unresolved issue. The count is useful only when paired with context.
Unresolved ownership may be visible when no role is recorded as responsible for the next step, when several roles each wait for another to act, or when the record repeatedly says a decision is pending without identifying who must advance it. A named owner is not enough if that role lacks the authority to obtain a decision or escalate the impasse. Conversely, a clearly authorized body can still face a capacity bottleneck if its members or staff cannot review the workload in a timely way.
Worked Example: Interpreting Repeated Referrals
Hypothetical data. A proposed eligibility change is submitted on 2 April. On 5 April, the sponsor refers it to the platform committee. On 12 April, the committee returns it to the sponsor, asking who can approve the shared-population wording. On 18 April, the sponsor sends the same question to the coordinating center. On 25 April, the coordinating center returns it to the committee, noting that the committee's approval is still required. The committee records a final decision on 30 April.
The elapsed time is \(30 - 2 = 28\) calendar days. The same authority question moved from committee to sponsor and then to coordinating center before returning to the committee. This is one identifiable referral loop in the decision path, not simply three planned consultations: the dated notes show the same approval question remained unresolved.
The pattern supports an authority or ownership diagnosis more strongly than a pure capacity diagnosis. The key evidence is not the 28-day duration; it is that the roles returned the same question without identifying a role empowered and accountable to settle it. However, the records alone do not prove that workload played no part. A complete diagnosis checks whether meeting availability or staff capacity also contributed and whether the applicable agreement already assigns the decision right.
Distinguish Authority Problems From Resource Constraints
The distinction is best made by comparing the timeline with the assigned roles and the actual work queue. An authority problem is plausible when the process cannot identify who has the right to decide, who must integrate required input, or who can escalate a disagreement. A resource constraint is plausible when authority and responsibility are clear, but the responsible people or body cannot complete the work promptly because available capacity is insufficient.
Useful evidence includes the documented decision route, the reasons for referrals, meeting schedules, queue size, reviewer availability, and whether comparable decisions wait at the same stage. No single indicator establishes the cause. For example, a long queue could reflect understaffing, unusually complex cases, or a deliberate limit on review frequency. Repeated requests for the same clarification can arise from unclear authority, but they can also indicate that the submission itself is incomplete.
| Observed pattern | What it may indicate | What to verify |
|---|---|---|
| Same question returns among roles, with no clear final decision-maker | Unclear authority or unresolved ownership | Formal decision rights, escalation route, and who must integrate advice |
| One authorized body owns the decision, but cases wait in a queue | Possible capacity or scheduling constraint | Workload, meeting cadence, available reviewers, and queue conditions |
| Repeated requests for missing or inconsistent material | Incomplete submissions or unclear process requirements | Submission criteria and whether requirements were communicated in advance |
| Several distinct reviews occur in a planned sequence | Potentially appropriate process, not necessarily a bottleneck | Whether each review has a defined purpose and the sequence is required |
Worked Example: Comparing Authority and Capacity Evidence
Hypothetical data. Two proposed document changes each enter the same governance route. For Change A, the authorized committee is identified in the agreement, the owner is recorded, and the committee considers the change at its next scheduled meeting 21 days later. The record shows no referral or dispute. For Change B, the committee meets within 3 days, but returns the proposal twice because the sponsor and coordinating center disagree about which role can approve the change; the final decision is recorded 20 days after submission.
For Change A, the timeline shows a 21-day wait to the next meeting, while authority and ownership are documented. That is consistent with a scheduling or capacity constraint, though more evidence is needed to tell whether the cadence was avoidable or appropriate. For Change B, the total is \(20\) days, and two returns occur despite a meeting within 3 days. The repeated authority dispute points to unclear decision rights as a more immediate explanation than a shortage of meeting opportunities.
The comparison illustrates why total duration is not a diagnostic classification. Change A took one day longer, but its documented path differs. A defensible conclusion is that Change A warrants checking the meeting queue and available capacity, while Change B warrants clarifying authority and ownership. Neither conclusion rules out a secondary contribution from the other type of constraint.
A Practical Diagnostic Sequence
Name the decision, define the start and endpoint, and state the time unit. Separate one decision from a broader amendment containing several distinct decisions.
Order submission, review, referrals, responses, meetings, and resolution. Mark gaps and uncertain intervals as uncertain rather than assigning them an assumed cause.
Separate active and waiting intervals where evidence allows. Identify repeated referrals of the same unresolved question and the stages at which they occur.
Compare the observed path with assigned decision rights and ownership. Then check meeting access, queue, reviewer availability, and workload before attributing delay to resources.
Separate observed facts from interpretation, name plausible competing explanations, and identify what additional evidence would distinguish them.
Justifying the Diagnosis
A timeline can be understood as a sequence of non-overlapping intervals. If the intervals cover the full period from start to decision, adding them gives the total elapsed time:
Here, \(t_j\) is the duration of interval \(j\), and \(k\) is the number of intervals. This identity justifies locating where time accumulated, provided the intervals share the same unit, cover the whole period, and neither overlap nor omit time. It does not prove that the longest interval caused the delay: an interval may be long because the governing process deliberately requires a review period, and another earlier problem may have caused the wait.
Similarly, a repeated referral establishes a sequence of events only when the records show that the same unresolved question was returned. It supports an authority diagnosis when that sequence coincides with uncertainty about who can decide or escalate. It supports a resource diagnosis when responsibility is stable and documented, while evidence shows the assigned decision-maker is waiting on limited review capacity. These are diagnostic arguments from converging evidence, not automatic conclusions from a numerical threshold. There is no universal number of days or referrals that defines a governance bottleneck across all platforms.
Failure Modes, Alternatives, and Limits
A common mistake is to see the longest interval in a timeline and conclude that it must be the bottleneck. For example, a 14-day interval before a meeting may look like a capacity failure. The correction is to ask what the interval represents, whether it is a normal and necessary review cadence, whether the decision was ready for that meeting, and whether the assigned body had the authority to resolve it. Duration locates time; contextual evidence supports a diagnosis.
Another mistake is to treat every referral as evidence of confusion. Planned independent or specialist review may be a necessary safeguard. The relevant comparison is not “one referral versus none,” but a defined review route versus repeated circulation of the same unresolved issue. A further failure is to infer adequate capacity merely because a decision owner is named. An owner can be accountable yet unable to act promptly with the available resources.
The main alternatives are to improve role clarity, address resource capacity, or retain the process while improving submission completeness or scheduling. These responses should follow the evidence rather than precede it. If several explanations remain plausible, describe a mixed or indeterminate diagnosis and identify the records needed to resolve it. Incomplete timelines, informal communications outside the record, changing decision scope, and overlapping delays all limit certainty. The diagnostic result should be a reasoned account of the process, not a performance score or a claim that the governance system caused an outcome in the trial.
Check Your Understanding
Answer each question using the diagnostic distinctions in this tutorial. Short solutions follow the questions.
- A decision takes 18 calendar days from a complete submission to approval. What additional information is needed before calling it a resource bottleneck?
- A decision is returned twice because the same two roles each state that the other must approve it. Which diagnosis is most directly supported, and what should be checked?
- A timeline has 5 days of active review, 9 days waiting for a meeting, and 4 days of active follow-up. Calculate total elapsed time and the waiting share.
- Why should planned sequential reviews not automatically be counted as repeated referrals?
- A role is clearly authorized and owns the decision, but its review queue has grown while reviewer availability has fallen. What is the leading explanation, and what limitation remains?
Solutions. (1) Check the defined start and endpoint, the stages where time accumulated, whether the process required the interval, the meeting queue, available reviewers, and whether authority was clear. The 18-day total alone is insufficient. (2) Unclear authority or ownership is most directly supported. Check the formal decision rights and escalation route, and determine whether capacity or incomplete information also contributed. (3) Total elapsed time is \(5 + 9 + 4 = 18\) days. The waiting share is \(9/18 = 0.5\), or 50%. (4) Distinct planned reviews have different purposes and may be required; a repeated referral concerns the same unresolved question returning without resolution. (5) A resource constraint is the leading explanation because authority is stable while queue and availability evidence point to limited capacity. The evidence does not rule out other causes, such as unusually complex cases or an appropriate review schedule.