Out-of-specification paracetamol batch: two assays disagree and the warehouse wants release

Yes. The HPLC assay of 92.4% is outside 95.0-105.0%, so the batch has produced an OOS result and the OOS procedure is triggered. The fact that a second method gave a passing figure does not cancel the first result.

What must happen, in order:

1. No retesting, no release, no destruction until the investigation is documented. Retesting the same composite until a passing figure appears is testing into compliance - it is data manipulation, not analysis.
2. Phase I laboratory investigation - check system suitability records, standard preparation and purity, weighings, dilutions, volumetric glassware, column history, integration, and the calculation. Record the documented deviation (30-minute sonication against 10 minutes in the validated method) and determine whether it could affect the result.
3. Phase II manufacturing investigation only after the laboratory is shown to be blameless, or in parallel if the physical data already suggest a product problem.
4. Written hypothesis and a pre-approved retest plan - how many replicates, from which sample, by which method, with acceptance criteria fixed before the results are seen.

Here the physical data already point away from a purely laboratory explanation: hardness 4.0 kp against a 6-9 kp specification, friability 1.5% against NMT 1.0%, and dissolution 74% against Q = 80%. Two independent, non-laboratory failures exist. That means the batch has a real quality problem regardless of how the assay dispute is resolved, and the OOS investigation should be run on the assumption that the product may be defective.

The HPLC result should drive the decision.

• The HPLC method is the validated, stability-indicating method used for registration; it separates the analyte from degradants and excipients. The UV method at 243 nm is a single-point, non-specific measurement that sums everything absorbing at that wavelength - excipient absorbance (talc, stearate-derived material, any residual solvent) and standard-preparation error both inflate it. A 6% absolute gap is far too large to be explained by placebo interference, which points to an error in the UV standard or dilution.
• Degradation cannot explain the low HPLC figure, and the 4-aminophenol data prove it. Paracetamol hydrolysis yields 4-aminophenol. Losing 8% of a 500 mg label claim means roughly 40 mg of paracetamol converted; on a molecular-weight basis (151.16 to 109.13) that would generate about 29 mg of 4-aminophenol per tablet, i.e. around 5.8% of label claim. The measured value is 0.03% - nearly two hundred times lower. So 30 minutes of sonication did not destroy 8% of the drug; the loss is either arithmetic or genuine low content.
• That cuts both ways: the laboratory must now prove its own error, because the alternative is a subpotent batch. The burden is not on the product to produce a passing number.

Meanwhile the hardness, friability and dissolution failures are independent of any assay argument, and dissolution failure is the clinically relevant one - a tablet that releases 74% at 45 minutes will under-treat a patient even if the label claim is correct. Do not change the dissolution specification to accommodate the batch.

Decision: reject the batch. Do not release the 150,000 tablets against the distributor order.

Grounds: dissolution below Q; hardness below specification; friability above specification; and an unresolved OOS assay with a 6% method discrepancy. Any one of the physical failures would justify withholding release; together they indicate a batch that is mechanically weak and biopharmaceutically inadequate.

Technical CAPA, addressing the most probable root causes:

1. API particle size. Coarser API (d50 85 versus 30 micrometres) reduces surface area and dissolution rate; this is the cleanest explanation for the dissolution failure with an unchanged formula. Add a particle-size specification to the API monograph and to supplier qualification, and re-qualify the new source before further use.
2. Compression/granulation. Low hardness with high friability points to inadequate binder, over-dried granules, low compression force, or excessive turret speed. Review granulation end-point moisture, binder concentration and compression parameters, then demonstrate capability on three consecutive validation batches.
3. Analytical. Retrain on the validated sonication time; repeat the Phase I exercise; add a second analyst check on dilutions for assay.

Regulatory: document the failure and the investigation in the batch record; quarantine the batch under lock; destroy it under DDA-approved, witnessed procedures with a destruction certificate; if any part of the batch has already left the site, initiate a recall, classify it according to health hazard, and notify DDA in writing. Also review whether the change of API supplier required a variation to the registration - an unapproved source change is a separate compliance finding.