The conclusion first: when new evidence bears on an earlier Quality decision, the first task is a judgment, not a filing step. Either the original basis still holds and the record should say why, or it no longer holds and the decision needs to be made again by the authority that owns it.
What the regulations require, and what they leave to judgment.
For incoming components used in finished pharmaceuticals, 21 CFR 211.84 requires each lot to be withheld from use until it has been sampled, tested or examined, and released by the quality control unit (211.84(a)). Representative samples must be taken of each shipment of each lot, with the number of containers sampled and the amount taken based on appropriate criteria, such as statistical criteria for component variability, confidence levels, the precision desired, the supplier's past quality history, and the quantity needed for analysis (211.84(b)). At least one test to verify identity is required (211.84(d)(1)). A supplier's report of analysis may be accepted in place of testing each component against all appropriate written specifications only if at least one specific identity test is performed and the manufacturer establishes the reliability of the supplier's analyses through appropriate validation of the supplier's test results at appropriate intervals (211.84(d)(2)).
21 CFR 211.160 requires specifications, standards, sampling plans, and test procedures, including any change to them, to be drafted by the appropriate organizational unit and reviewed and approved by the quality control unit (211.160(a)). It requires laboratory controls to include scientifically sound and appropriate specifications, standards, sampling plans, and test procedures (211.160(b)).
Read together, those sections tie sampling to stated criteria, including supplier history, and make the sampling plan a quality-unit decision. They do not set a review interval for a sampling plan. They do not say which new information makes an earlier plan inadequate, and they do not use the term “risk assessment.” Those questions are left to the firm's documented judgment.
Guidance supplies useful, nonbinding context. ICH Q9(R1) Quality Risk Management, adopted by FDA as guidance in May 2023, describes risk review as an ongoing part of quality management. It recommends reviewing the output of a risk assessment to take new knowledge and experience into account, and it says the level of effort, formality, and documentation should be commensurate with the level of risk.
Component control is also a recurring inspection topic. In FDA's FY2025 inspection observations data for drugs (October 2024 to September 2025), observations citing 21 CFR 211.84(d)(2) include accepting suppliers' reports of analysis without a specific identity test or without establishing the reliability of the supplier's analyses, and establishing that reliability without validating results at appropriate intervals. These are Form 483 observation counts, not findings about any particular firm, and they say nothing about how any inspection ended.
Separate a record update from a fresh decision.
New evidence usually lands in one of three places.
It confirms the original reasoning. The decision already considered this kind of risk, and the new information falls inside what was weighed. The record should say so, citing the part of the original assessment that already addressed it. That is a record update.
It adds detail without changing the conclusion. The information is relevant, but the evidence the decision depended on is still intact. The record should show the review and explain why the conclusion stands. That is still a record update, with judgment shown.
It removes or weakens part of the basis. An assumption no longer holds, a source of evidence is less reliable than believed, or a risk the decision did not consider is now visible. The original approval no longer explains the current state. Updating the file cannot fix that. The decision has to be made again, on the evidence now available, by whoever holds the authority.
The hard part is not the definitions. It is noticing which case you are in before the paperwork decides for you. A note to file can make a weakened basis look reviewed. A revised risk score can make a new decision look like routine upkeep.
An illustrative scenario
Both answers are possible, and the scenario does not decide which is right. What matters is which question the record answers.
If the January assessment relied on the supplier's reports of analysis and treated them as reliable because of the supplier's qualification history, the warning letter may bear directly on that assumption, even though it concerns another firm's testing. If the January plan already included the manufacturer's own testing for the attribute the letter cited, and the reliability of the supplier's reports was checked at defined intervals, the letter may confirm a risk the plan already covered.
The quality director should be able to point to the line in the January record that settles it. If that line exists, the response is a documented review that cites it. If it does not, the original decision did not address the risk now visible, and filing the warning letter does not make it address that risk.
Use the Four Lenses to test whether the basis still holds.
The Four Lenses apply to the reassessment as much as to the original decision. Each asks a different question about whether the earlier approval still explains what the site is doing today.
Evidence
What did the original decision rely on, and what is known now? List the evidence the decision depended on, such as supplier qualification data, reports of analysis, in-house test history, and the risk assessment. Set it beside the new information and its source and date.
Judgment
Does the new information fall inside what was already weighed? Show the reasoning that connects it to a specific part of the original basis, and say whether it confirms, refines, or undermines that part.
Risk
What uncertainty does the site carry now that it did not carry before? Name the exposure while the question is open, any interim control, and the point at which the open question must be escalated.
A practical record example
Applied to the illustrative scenario, a reassessment record might look like the table below. The content is fictional and shows the structure of the record. It does not determine the right outcome for a real site.
| Lens | What the record contains | Question the record answers |
|---|---|---|
| Evidence | January sampling plan and risk assessment; the supplier qualification data and reports of analysis it relied on; in-house identity and other test results since January; the June warning letter, its date, and the excipient class it cited. | What did the January decision depend on, and what new information bears on it? |
| Judgment | Comparison of the cited testing gap with the January plan: which attributes the site tests itself, which rely on supplier reports, when supplier results were last checked, and whether the January assessment considered this failure mode. | Does the warning letter fall inside what was weighed in January, or outside it? |
| Risk | Open exposure while the review is under way; any interim control the site chooses, such as holding lots pending added testing; owner and date for closing the review. | What uncertainty exists now, and how is it being managed until the question is settled? |
| Authorization | Either a documented confirmation that the January approval stands, citing the part of the original assessment that already covered the risk, or a new quality-unit decision on a revised plan that references and replaces the January approval. | Is the current sampling plan explained by a decision someone with authority actually made? |
Three questions that test whether evidence has changed the basis.
- What did the decision depend on? Name the specific evidence and assumptions without which the original conclusion would not follow.
- Does the new information touch any of them? Connect it to a specific dependency, or record why it does not touch one.
- If it does, who decides again? Identify the authority that owns the decision, the evidence available now, and how the new decision relates to the earlier one.
The executive conclusion
Decisions age. The evidence behind them can weaken while the approval stays signed. The Head of Quality's job is not to reopen every decision whenever something new appears. It is to make sure the organization can tell whether new evidence confirms an earlier decision or removes part of its basis, and that a decision whose basis has changed is made again by the authority that owns it rather than amended into place.
For more on what makes a decision record explainable, read The Record Should Explain the Decision and What evidence did this decision rely on?
Source context: Regulatory requirements: 21 CFR 211.84 and 21 CFR 211.160. Nonbinding guidance: FDA/ICH, Q9(R1) Quality Risk Management, May 2023. Inspection data: FDA, Inspection Observations, FY2025 drugs data. The excipient scenario, record example, and Four Lenses review are fictional CW editorial examples, not regulatory mandates.
