Sandoz co-amoxiclav SmPC error: when the label, not the batch, is the defect

What happened

On 5 November 2026 the MHRA issued a Class 4 medicines defect notification for Sandoz Limited's co-amoxiclav powder for oral suspension, 250/62.5 mg/5 mL. The dosing table in the product's Summary of Product Characteristics (SmPC) was found to be wrong. According to the alert, the error could double the weight-based dose. All batches are affected.

The records do not say which weight bands or age groups the table covers, what the correct values are, or how the error came about. They also do not say whether the patient information leaflet or the packaging carries the same table, or whether any patient was harmed. Nor do they describe any corrective action by Sandoz or the MHRA beyond the notification itself. Those gaps matter, and readers should check the full notice before acting.

Why a Class 4 notice still deserves attention

The MHRA grades defect notifications from Class 1 (most urgent) to Class 4. Class 4 is the lowest grade and is generally described as 'caution in use'. It is not a recall. Class 4 notices often cover problems that do not call for removing stock from the market.

Here the grade needs careful reading. The defect is in information, not in the product's composition or manufacture, so the tablets-and-vials logic of recall classification fits poorly. At the same time, the notification flags a possible doubling of a weight-based dose in a medicine used by weight. The records do not explain the MHRA's reasoning for the classification, so we should not speculate about it. The practical point is that a low-numbered class is not a reliable proxy for clinical consequence when the defect sits in dosing instructions.

Because every batch is affected, there is no clean batch-based segregation to apply. Anyone relying on the published table is exposed, whatever lot is on the shelf.

The quality-system lesson

A wrong SmPC table is a failure in labelling and document control. It does not involve a manufacturing deviation. That makes it a useful test of whether a quality system treats product information with the same rigour as batch records. Typical weak points include transcription between source documents, review of numerical tables by someone other than the author, and reconciliation of the SmPC against leaflet and pack text after each variation.

The records do not identify the root cause, so none should be assumed. The case is still a reason for QA teams to ask whether dose tables in their own product information have been independently verified against the approved source, and whether that check is recorded.

Who is affected, and what we do not yet know

Direct exposure sits with prescribers, pharmacists and carers using the Sandoz 250/62.5 mg/5 mL suspension in the UK. Sandoz's regulatory and quality functions, and its QP, will be handling the correction. Wholesalers and dispensing systems that copy SmPC dosing text into clinical decision tools are also exposed, because an error in the source document can spread to downstream references.

Still unknown from the records: the date the error was introduced, how long the table was in circulation, how many prescriptions or doses it touched, and whether a corrected SmPC has been published.

Context: other notices published the same week

Several items published on 6 November show different faces of the same issue, which is how much a public notice actually tells the reader. BfArM posted urgent field safety notices for the Ion Torrent Genexus Dx software, the Helena Biosciences V8 UltraCE, Intuitive Surgical's S. Monopolar Curved Scissors Tip and exocad's exoplan software. In each case the public listing gives little or no detail, and the substance sits in the manufacturer's attached advisory. The V8 UltraCE listing, for example, gives no detail on the problem, affected lots or required actions.

Health Canada published Type II recalls for Sysmex CN-3000 and CN-6000 coagulation analyzers, where probe deposits risk reduced sample volumes, and for the Hettich ROTOFIX 32A centrifuge, where the lid lock may not engage securely. It also warned of bacterial contamination in unauthorised injectable peptides sold online and reported seizing over 11,400 vials. These are device and illegal-supply issues, not comparable defects to the Sandoz case. Together they underline that the first public notice is often a pointer, not a full account.

Separately, the US Department of Justice reported a $1.25m False Claims Act settlement with Lab Genomics LLC and owner Leena Dalal over volume-based marketer commissions allegedly disguised as flat fees. It is a compliance story, not a quality one, but it shares the theme that what documents say can differ from what actually happens.