Running a salmonella recall drill with pad lot numbers

The drill starts at 10:05 on a Tuesday with a fictional lot number and a question: if this lot of cooked chicken were implicated, which cases went out with a pad from the same delivery, and where are they now? The clock stops when the list is complete. In most plants the first attempt takes four to six hours. The second attempt, six months later, should take under two.

Pads are a small line item and an easy one to leave out of a traceability exercise, which is exactly why they turn up as the weak point. A recall does not care that the pad cost four pence. It cares whether you can say which production runs used it.

What the drill actually tests

Not whether you can find a supplier’s certificate. Whether you can go from a pad lot number to a production date, and from a production date to a customer list, in one working session. Three links, and the middle one is usually broken.

The break happens because pad lots get recorded the way goods-in records everything else: by delivery date, not by the code on the carton. If two pallets arrived in the same week from different batches and were put away side by side, the goods-in line reads the same for both. When you need to separate them, the paperwork cannot.

Supplier coding practice makes this worse. Some pad makers print a lot code on the carton and nothing on the inner bag; others print a code that changes monthly rather than per production run. Ask for the code structure in writing before the first pallet arrives, and check that the code on the carton actually changes when the batch changes. A code that has not moved in nine months tells you nothing.

Building the lot map before you need it

The fix is a single column in the production record. Every run gets the pad lot code against it, written by the person who fitted the roll rather than by the office the next morning. That is a two-second job at the line and an hour of arguing saved at the other end.

• Record the code printed on the carton, not the delivery note number.
• Keep part-used cartons with their code visible; never tip sheets into a plain dispenser.
• If two lots are open at once, mark the changeover time in the production log.
• Photograph the carton label when the pallet is opened. One photo per pallet is enough.

With that in place, the drill becomes mechanical. Lot to date, date to run, run to pallet, pallet to customer. The only judgement call is whether the pad was in contact with product or sealed inside a pack, and that should already be in the pack specification.

The four-hour failure and what it cost

A ready-meal plant we know ran its first mock recall and stopped at the pad link for most of the morning. One pallet had been split across two lines, the cartons had been decanted into a dispenser, and the code had gone out with the cardboard. The exercise ended with a note that said the pad could not be traced, which in a real event means recalling wider than necessary.

Recalling wider is expensive but safe. The dangerous version is the opposite: assuming the pad is irrelevant because it never touches the food, and leaving it out of the traceability scope altogether. That assumption holds for a sealed pack and fails for a tray of loose portions sitting directly on the sheet.

Run the drill twice a year, alternate between a raw and a cooked product, and include one supplier-side question: can your pad supplier give you a lot list for a given month within a working day? If the answer is no, the drill will fail at the first link no matter how good your own records are. The pad documentation guide lists what a supplier should be able to produce.