Holding time: the study sets the limit; the routine is where it fails

A holding-time study answers a useful question: how long an intermediate or bulk product stays within specification under the plant's real storage conditions. The question that usually decides risk is different — whether that limit is still being met on the shop floor, batch by batch, when production slips, the portfolio grows, and the operator has more than one limit to track at once.

RDC 658/2022, article 274, covers the stability of intermediates and bulk stored for prolonged periods before packaging or shipment — a narrower slice than every inter-stage hold on the shop floor. The market uses "holding time" more broadly; for study design, plants often also anchor to WHO-style guidance. The regulation requires the impact to be assessed; it does not set a universal number or prescribe a step-by-step method. The study justifies the limit. Routine control is what turns that limit into evidence.

So it starts with what the study has to settle.

What the study must decide

A holding-time protocol is not a generic worst-case exercise. Different formulations age differently, and the design has to mirror the real process — not a laboratory scenario the plant never runs.

The points a protocol usually locks before the first sample:

• Number of batches and a sampling plan with enough material for every intermediate time point and for cumulative time across phases. One batch can be acceptable when risk analysis supports the conclusion; in many cases, using a full batch is what makes the evidence defensible.
• The maximum realistic wait in routine until the next step — not a comfortable number on paper, but what operations can actually generate.
• Location, storage condition and container type identical to routine, preferably under temperature and humidity control.
• Tests driven by the physicochemical changes plausible in the interval; microbiological assessment according to product/process risk — and, when the protocol requires it, at the defined time points.
• Acceptance criteria — which may be tighter than the specification used for batch approval, depending on risk.
• Validated analytical methods and, where possible, stability-indicating ones.

The final report closes with a critical evaluation of the results and with the applicable limit between phases. It also states whether the limit enters routine immediately or waits for ongoing stability results. Without that decision, the study stays an approved document; the plant keeps operating without a rule.

Holding time without routine control is a number in a report. It is not a control.

Where the risk actually sits

Once the limit is set, the problem moves: every manufactured batch needs to show that the established limit was respected. Exceeding holding time is a quality deviation: it requires investigation, risk assessment and, often, placing the batch on ongoing stability before any batch approval decision for commercial supply.

The hard part is not remembering that principle in training. It is applying it when there are dozens of products, different times between steps, overlapping shifts and waits that do not fit a spreadsheet updated "when someone has time." Without an automatic mechanism, effective control stays fragile — and GMP fragility shows up early in a health authority inspection.

Silent-failure signals usually arrive before anyone admits the problem:

• the limit sits in an approved report, but the batch record never shows the clock starting and stopping;
• the operator calculates by hand or writes on a side sheet;
• prolonged waits become deviations only if someone notices later;
• intermediate sampling times from the study never became alarms in routine;
• products with different limits share one generic spreadsheet.

In those cases the study did its job. The execution system did not.

Enforce the limit at the moment of control

If the gap is in execution, so is the fix. Enforcing holding time in routine takes three system capabilities — and none of them is another dashboard:

1. The clock starts on the right event — end of step, intermediate approval, entry into the holding container — with an auditable timestamp, not a late keyboard entry.
2. The limit travels with the material — by product, by step, by storage condition, versioned under change control, not editable "in place" by whoever is in a hurry.
3. An exceedance becomes a block or an exception — with a trail of what was evaluated, against what and when — before the next step consumes the material.

That is the same shift in the moment of control that a well-designed electronic batch record aims for: from review after the fact to execution while the batch is still happening. In plants that use NEO EBR, the logic is the same — the system only helps if the limit sits in the flow, not if the deadline exists only in the study PDF.

What the study and the software do not replace

It is worth being clear about where that automation stops. No system creates a defensible holding time if the sampling protocol did not reflect routine, if the study container was not the plant's, or if acceptance criteria were copied from the specification used for batch approval without looking at interval risk. Software also does not replace the quality decision on whether to approve a batch that exceeded the limit — it only makes the overrun visible in time.

Data integrity still applies: if the start of the wait is typed hours later, the control inherits the weakness of that transcription. The same is true of limits that are too loose to ever fire, or so tight that operations learn to work around the alarm.

Questions before declaring HT under control

Putting study and routine together, these questions usually separate an approved document from an effective control:

• Does article 274 of RDC 658/2022 sit on a study that used the plant's real storage conditions?
• Is each approved limit on the execution record for the corresponding batch, or only in the report?
• Can the operator follow the limit without side calculation or a parallel spreadsheet?
• Does an exceedance generate a deviation with evidence — or does it depend on someone remembering to look?
• Are different times and products versioned, or forced into a generic rule too coarse to be true?
• Has Quality agreed in writing on what can be approved, without a manual re-read of the clock, when a batch is within HT?

The last question is usually the most deferred — and the one that surfaces when an inspection asks for the link between study and batch.

Closing the loop

Good holding time is not an approved PDF. It is a limit the study justifies and that routine can enforce, evidence and investigate when it fails. The study sets the limit; the routine is where it fails — or where it finally becomes a control.

In your plant, does holding time stop the next step in time — or does it still live only in the report?