Why Timing Matters
A model discussion is useful only if its timing fits both the state of the evidence and the decisions ahead. Discussing a model before the relevant data, assumptions, or proposed use are sufficiently clear can make the exchange premature. Waiting until a protocol, dose strategy, or trial design is effectively fixed can make otherwise useful feedback difficult to implement.
The planning problem is therefore not simply “early or late.” It is to identify the development decision the discussion may inform, estimate when that decision must be made, and allow enough time to prepare, interact with the agency, evaluate the feedback, and make any resulting changes. The earlier material on separating uncertainty that can change the decision is useful here: timing is especially important when unresolved model uncertainty could alter a consequential choice.
Milestones such as candidate selection, first-in-human (FIH) planning, dose selection, and confirmatory-trial design are practical anchors for this analysis. They are not interchangeable. Each has different evidence needs and different consequences if the underlying assumptions change. A model intended to compare candidate compounds may need to be discussed before a candidate is nominated; a model informing a confirmatory trial may depend on clinical evidence that does not yet exist at nomination.
A Backward-Planning Framework
Start from the point by which the development team needs to act, rather than from a preferred meeting date. Estimate how long it will take to prepare a decision-ready package, complete the selected interaction process, and respond internally to the feedback. Include a reserve for uncertainty in those estimates. These durations are planning assumptions, not guaranteed agency timelines.
This relationship follows from adding the periods that must fit before the decision date: preparation precedes the interaction, and time to assess feedback and make changes follows it. If any period overlaps with another, the plan should represent that overlap explicitly rather than subtracting it twice. If the calculated initiation date has already passed, the team should not pretend the original schedule remains feasible; it should consider moving the decision, narrowing the intended use of the discussion, or proceeding with a clearly stated plan for managing the unresolved issue.
In practice, “preparation time” includes generating or assembling the evidence needed for a meaningful discussion. “Interaction time” is the expected time from the relevant submission or request to the exchange, based on the chosen route and current planning information. “Response time” is the time needed to interpret feedback, assess its implications, and revise analyses or plans. The reserve reflects uncertainty in these estimates. A route’s actual procedures and scheduling constraints must be checked separately; the arithmetic is a planning aid, not a claim about a standard agency service time.
Specify the decision date or the period when a choice must be made, and identify what becomes difficult to change afterward.
Separate evidence that must be available for a useful discussion from evidence that can reasonably be generated later.
Allow for package preparation, the selected interaction route, internal evaluation of feedback, and a realistic reserve.
Check whether the team still has practical time and authority to alter the model, analysis, dose plan, or trial design if the discussion changes its assessment.
Worked Examples Across Development Milestones
Worked Example: Candidate Selection
A hypothetical team plans to nominate one of two candidate compounds at week 20. A model comparing projected exposure and a prespecified pharmacological target is relevant to that choice. The team estimates 4 weeks to assemble the evidence and prepare a suitable package, 9 weeks for the selected interaction process, 3 weeks to evaluate feedback and revise its comparison, and a 2-week reserve. These are internal planning estimates, not standard regulatory timelines.
Check the schedule. The total time required before nomination is \(4+9+3+2=18\) weeks. Working backward from week 20 gives \(20-18=2\): the latest planned initiation is week 2.
Interpretation. If the team starts at week 5, the estimated cycle ends at week \(5+18=23\), three weeks after the nomination date. The discussion might still help with later development, but it cannot reliably inform the scheduled candidate choice unless the decision is delayed or parts of the cycle are shortened without undermining its purpose. The team should also check whether the comparison model and its inputs are sufficiently mature by the time the package is prepared; a timely meeting with inadequate evidence is not a successful timing plan.
Worked Example: First-in-Human Planning
A hypothetical sponsor expects to lock key FIH protocol assumptions in 15 weeks. It estimates 5 weeks to finish the relevant model analyses and prepare the package, 8 weeks for the planned interaction process, 3 weeks for internal review and possible revisions, and a 2-week reserve.
Calculate the required lead time. The total is \(5+8+3+2=18\) weeks. Therefore, the latest initiation date from this plan is \(15-18=-3\) weeks relative to today: the schedule is already three weeks short.
Assess the options. The team could move the protocol-lock date, if feasible; seek an earlier interaction based on a narrower but sufficiently supported aspect of the model plan; or proceed while documenting which model uncertainties remain and how they will be addressed. It should not count feedback received after protocol lock as though it had informed the initial protocol assumptions. Nor should it rush an analysis whose evidence is not yet fit for the intended discussion merely to meet an arbitrary calendar date.
Worked Example: Dose Selection and Confirmatory-Trial Design
A hypothetical program has an important dose-selection decision in 14 weeks. The planned model update depends on an interim data set expected in 2 weeks. After that readout, the team estimates 4 weeks for analysis and package preparation, 7 weeks for the selected interaction process, 3 weeks for internal assessment, and a 2-week reserve. The decision date is fixed for this exercise.
Calculate from evidence availability. The earliest planned initiation after the readout is week 2. The remaining cycle requires \(4+7+3+2=16\) weeks, so completion is estimated at week \(2+16=18\), four weeks after dose selection.
Interpretation and adjustment. Waiting for the interim readout makes the complete planned cycle too late to influence the dose decision. The team could discuss an earlier, adequately supported model plan before the readout and later update its assessment, or reconsider whether the dose decision can move. If the program’s next relevant choice is confirmatory-trial design, it may instead plan the fuller discussion before that design is locked, provided the later decision remains genuinely open and the evidence available at that time supports the discussion. These are alternatives to assess, not guarantees that a particular route will be available or that an agency will endorse the model.
Matching Timing to Evidence and Milestone
Different milestones create different timing pressures. Candidate selection often requires comparing options while there is still a feasible opportunity to choose among them. FIH planning requires a useful discussion early enough to influence the planned approach, while recognizing that some uncertainties may remain because clinical data are not yet available. Dose selection may depend on accumulating exposure and response information; waiting for a more informative data cut can improve the evidence but consume the time needed to act on feedback. Confirmatory-trial design may allow a longer evidence-development period, yet a discussion after key design features are locked may have limited practical influence.
The appropriate balance is not “discuss every model as early as possible.” It is to discuss at the earliest point when the evidence is sufficiently interpretable for the intended purpose and the decision remains open to change. A staged approach can be useful when evidence will mature in steps: an earlier discussion can address a supported modeling plan or assumptions, followed by later work that uses new data. Staging has costs, including staff time and possible changes in the evidence base, so it should be justified by distinct decisions or meaningful changes in what can be assessed.
| Timing approach | Potential advantage | Main risk |
|---|---|---|
| Early, before key evidence is available | May leave more time to influence a decision or plan | The discussion may be premature if assumptions or evidence are too undeveloped |
| After relevant evidence matures, before the decision is fixed | Balances interpretability with an opportunity to act | Requires careful backward planning and realistic lead-time estimates |
| Late, after the decision is effectively fixed | May inform later decisions or clarify future evidence needs | Feedback may arrive too late to change the milestone that motivated the interaction |
Diagnostics, Failure Modes, and Limits
A timing plan is questionable if it lists a meeting date but not the decision date it is meant to inform; if it assumes feedback can be implemented without time for analysis or revision; or if the evidence required for the discussion will not exist until after the proposed interaction. Another warning sign is treating a calendar milestone as fully reversible when downstream work, manufacturing, protocol commitments, or other operational dependencies make a change impractical.
Review the plan when evidence-generation dates shift, the development milestone moves, or the expected interaction route changes. A simple schedule should show the dependencies, not merely a sequence of dates. Sensitivity checks help: recalculate the latest initiation date with a longer interaction estimate, a delayed data cut, or an additional revision cycle. If a small plausible delay makes the plan fail, the schedule has little resilience and the reserve may be inadequate.
A common mistake is to reason, “Earlier is always better, so we should request advice before the model is developed.” Earlier timing can preserve room to act, but without a sufficiently clear intended use and supporting evidence, the discussion may not address the decision the team needs to make. The correction is to distinguish early enough to change the decision from so early that the relevant issue cannot yet be assessed, then plan around both evidence readiness and remaining decision flexibility.
Timing analysis has limits. Planning estimates are uncertain; regulatory processes and meeting availability can vary; and a well-timed interaction does not ensure agreement, a particular outcome, or acceptance of a model. In addition, feedback may reveal a need for new analyses or evidence that cannot be generated before the original milestone. The team should use timing plans to expose these dependencies and make explicit contingencies, not to imply that an interaction removes scientific or operational uncertainty.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A team’s decision is in 12 weeks. It estimates 3 weeks for preparation, 5 weeks for the interaction, 2 weeks for internal response, and a 1-week reserve. What is the latest planned initiation date?
- Why might a discussion scheduled after a confirmatory-trial protocol is locked fail to inform the original design decision?
- A dose-selection model will be updated after a data readout in 4 weeks. The remaining preparation, interaction, response, and reserve periods total 13 weeks. A dose decision is planned for week 15. Does the full planned cycle fit?
- Give one reason an early discussion could be premature and one reason a late discussion could be ineffective.
- What should a team do if a plausible delay in the data cut makes its timing plan miss the decision window?
Solutions. (1) The required lead time is \(3+5+2+1=11\) weeks, so the latest initiation is at week \(12-11=1\). (2) The design may no longer be practically changeable, leaving insufficient time to implement feedback; the discussion could still inform later decisions, but not necessarily the locked design. (3) No. If initiation follows the readout at week 4, the cycle ends at week \(4+13=17\), two weeks after the decision at week 15. The team should reassess the decision date, the timing or scope of the interaction, or the plan for managing the unresolved issue. (4) An early discussion may precede interpretable evidence or a sufficiently developed model use; a late one may leave no feasible time to revise a consequential decision. (5) Recalculate the schedule, make dependencies and consequences explicit, and consider feasible contingencies such as moving the decision or using a suitably supported staged approach. It should not assume that the original timing remains achievable without adjustment.