Nine Farm Lots, One Wash Line, and a 24-Hour Clock

Genuinely missing:

• Harvest date and field or growing location for 3 of the 9 farm lots. These are the kind of key data elements a 24-hour traceback depends on, because harvest date and field are how you and an investigator narrow a produce event. Without them those three lots are dead ends, and in a real outbreak they would either be implicated by default or force a broader recall.
• A defined traceability lot code for the finished product. A Julian date plus line number is a production code, not a lot code that links to specific incoming lots. It cannot be resolved to a supplier lot, a quantity, or a location.

Exists but is not retrievable in the window:

• Receiving data in scanned PDFs. It exists, but retrieving it means opening files and reading them by hand. That is what consumed your four hours, and a 24-hour requirement leaves no room for it. Records must be sortable, linked, and available electronically.
• The finished-to-raw transformation. Without a system link between the case code and the lots consumed, every traceback starts from a manual reconstruction.
• Mass balance. The 11% gap from unrecorded rework means the quantity data underlying every record is wrong. In a traceback, a mass balance that does not close is the difference between narrowing to 2 lots and recalling 9.

Framing for management: the failing is not that the plant lacks records. It is that the records are not designed to answer the question the regulation and a recall both ask: which raw lots are in this case, and where else did they go.

Define the traceability lot. The finished lot should be the smallest unit you can actually run and hold. Practically, that is a wash-line run window of a few hours, not a full shift, and not a single truckload of greens. Each finished lot code should be assigned at packaging and carry an explicit, recorded list of the raw lots consumed during its run window. The trade-off is real: shorter run windows mean more changeovers and more finished inventory in small lots, which increases holding cost and warehouse complexity. That cost has to be presented to operations with the recall-scope benefit attached, or it will be overruled.

Record at the point of use. Raw lot codes must be scanned at receiving and again at the wash line, not written on a clipboard and keyed in later. The receiving record must capture, per lot, the traceability lot code, the reference document, the location, the date, the quantity, and the supplier and growing location. Keep the paper trail as the source evidence but make the electronic record the operating system.

Handle rework explicitly. Rework from a prior day carries its own lot identity. Either it has been consumed within a defined window with its own traceability record linking it to the day it was made, or it is not allowed back into the line. The cleanest control is a rule with a number in it, such as rework from the previous shift only, recorded with its originating lot. Without that, every mass balance fails and every traceback widens.

Rehearse. Run a mock traceback quarterly with a 24-hour stopwatch, one step forward and one step back, and report the result to leadership with the gap identified. The first one will fail; that is the point of running it before an investigator asks.

With the current system: the positive case carries a Julian date and line number. You would have to treat everything packed that day as suspect, which is 1,500 cases at minimum and likely every romaine case on that line from that shift. Because rework from the previous day was fed back unrecorded, the prior day's production is also in question. Faced with that, most operations recall broadly and absorb the cost and the brand damage. The scoping is imprecise because the data is imprecise.

With the redesigned system: the case resolves to a run window and a defined raw lot list, say 3 of the 9 farm lots and a recorded quantity of prior-day rework. You can then go one step back to the specific growers and fields, giving the produce farm and the regulator something they can act on, and one step forward to which customers received the implicated run rather than the whole day. The recall narrows to what the data supports, and the justification is documented rather than asserted.

Two cautions. First, presumptive is not confirmed; hold and confirm before any public action, and use the hold to prepare the traceback package so the clock does not start cold. Second, do not let a narrow traceability result over-persuade you. If the wash water is the likely vehicle, the implicated set is the water, not the lot list, and every lot through that water is in scope regardless of what the lot codes say. Traceability narrows the question; it does not replace the hazard analysis. Write the scoping decision down with the reasoning, because the difference between a narrow recall and an overdue one will be judged on that document.