Out-of-Specification Assay and a Friday Shipment

The proposal is not acceptable. Averaging results across a second set of samples, chosen after the failure is known, is testing into compliance - it converts a failing result into a passing one by arithmetic. Testing a 'different bottle' also breaks the link to the original sample, so any conclusion about the batch is unsupported.

Correct sequence - Phase I laboratory investigation, before any retest:

1. Confirm the result was calculated and reported correctly - check the calculation, the dilution factor, the standard weight and purity, and whether the correct formula was used.
2. Check the analytical system - system suitability, peak purity, column performance and age, mobile phase preparation and pH, detector wavelength, integration parameters, autosampler vials.
3. Check the standard - the new lot opened that week is a red flag: was it stored correctly, dried as the certificate requires, within expiry, and is its assigned potency correct? Compare against a previously qualified standard.
4. Check equipment and technique - balance calibration and level, volumetric glassware, pipette calibration, analyst training record, whether the composite was prepared and extracted consistently with the validated method.
5. Document each hypothesis and its outcome in the OOS record.

Only if a confirmed, assignable laboratory cause is found may retesting proceed, under a pre-approved plan stating the number of retests, the acceptance criteria (all results within specification, RSD limit) and the sample to be used. If no assignable cause is found, the result stands.

The three results are tight (92.4–93.1%), which argues against an isolated analyst blunder and points toward a systematic cause.

Candidate causes and the data to pull:

• API potency and quantity dispensed. Supplier certificate of analysis, the actual potency used in the batch calculation, weighing records, and whether a potency adjustment was made. A 3% low assay often begins with an API assumption.
• Blend uniformity and content. In-process blend assay results, blend sampling points and times, blender speed and load, and whether the blend result itself was near the lower limit.
• Granulation and drying. Batch 2310 from the same granulation assayed 96.2% - a 3-point drop within one granulation is a trend, not noise. Review granulation end-point, water content, drying temperature and time, and any process deviation on 2311.
• Compression. Tablet weight and hardness data, weight variation, and compression force; a heavy tablet could mask a low-potency blend, a light one could lower the composite.
• Analytical method transfer. Whether the method's extraction is complete for a harder, more lubricated tablet - check dissolution results for the same batch.
• Trending. Plot assay results for the last 20 batches from the same product and line; a drift over several batches changes the picture from batch-specific to systemic.
• Retention samples. Test them as part of the assessment only under the investigation plan, not as a substitute for the original result.

CAPA follows from the confirmed cause: re-qualification of the supplier, blend validation, in-process control tightening, or method improvement.

The batch is rejected. A confirmed OOS result against the assay specification means it does not meet its registered specification and cannot be released. 'Only two points low' is not a category - the specification is the specification.

Immediate actions:

1. Formally reject the batch in the quality management system, with the OOS investigation report attached and the rationale for rejection recorded.
2. Physically segregate all 480,000 tablets into a locked reject/quarantine area, labelled, with stock records updated so no one can dispatch them by mistake.
3. Notify the distributor in writing that Friday's shipment will not be made and why, without disclosing confidential manufacturing detail. Do not offer a partial shipment or 'release on waiver'.
4. Notify the regulator as required by the applicable guidance for a batch failure, especially where the product is already in distribution anywhere.

Extent of the problem:

• Determine whether any part of batch 2311 has already left the site, and if so initiate a risk assessment for a recall.
• Re-examine batch 2310 and any other batches from the same granulation. If 2311 was 3 points below 2310, 2310 may be within specification but still part of a downward trend; testing retention samples of 2310 is justified by the trend, not by a fishing exercise.
• Check whether other products on the same line share the root cause.

Closure: disposition the batch (rework only if the registered process and specification permit it, otherwise destruction with witnessed records), complete CAPA, and verify effectiveness at the next batches. Do not release on the basis of a promise to investigate after shipment.