Salmonella-Positive Milk Powder, a Roof Leak and a Wet-Mopped Vacuum Line

No. Hold the 22,000 lb and initiate a withdrawal of the 8,000 lb.

Why the negatives do not carry the decision:

• Twenty samples of 375 g is 7.5 kg out of 22,000 lb (about 10,000 kg) - roughly 0.075% of the lot.
• Salmonella in a dry powder is typically present at very low level and is not uniformly distributed; it clumps in particles. If the lot carried, say, one organism per kilogram, a 7.5 kg sample set would have a low probability of catching anything, so a negative result is consistent both with a clean lot and with a contaminated one.
• The lot is presumptively contaminated before you ever test it: a component lot tested positive and went into the blend, and the plant has no kill step after blending. You do not need finished-product positives to establish that.

Actions:

1. Hold and segregate all 22,000 lb with the ingredient lot traceable to it; withdraw the 8,000 lb from distributors immediately rather than waiting for confirmation.
2. Notify the supplier of the nonfat dry milk and quarantine remaining stock of that lot and any other lots from the same supplier and period.
3. Convene the food safety team for a hazard and risk assessment, and make the disposition decision (destroy, or hold for a validated rework route only if one exists) on the record.
4. Use the negative results as supporting information in the record, not as the basis for release.

Root causes, in order of impact:

1. The water intrusion. The roof leak at the bag dump station put water into zone 1 of a dry process - the single most reliable way to create a Salmonella harborage in a low-moisture plant.
2. The wet-mopped vacuum transfer line. Water plus a mop introduced moisture into an enclosed conveyance that is hard to dry, and the line went back into service the same day. Any residue in a vacuum line is a perfect growth site.
3. No written dry-cleaning procedure and a wet clean only in March: cleaning is not defined, so behavior is ad hoc, and the plant has no standard for what "dry" means.
4. The Enterobacteriaceae cluster and the filler positive show the organism is established in the environment, not just in one ingredient lot.

Why not a full wet clean-down: flooding a dry plant drives contamination into cracks, welds, hollow legs and equipment internals and creates the moisture that lets it multiply; it typically spreads the problem and raises counts afterward. Use dry cleaning (scraping, vacuuming with HEPA filtration, wiping with approved low-moisture methods) as the default. Where wet cleaning is unavoidable, contain it to defined equipment, use a controlled procedure, verify dryness before restart, and document the moisture check.

Also fix: repair the roof and the drip path to the dump station, review the dump station's design and shielding, and add a water-intrusion response procedure tied to the EMP.

Design principles for a dry, no-kill-step plant:

Zoning. Define zone 1 (product contact surfaces: filler nozzles, blender internals, conveyor contact, scoops), zone 2 (adjacent non-contact surfaces that could transfer, including equipment frames, the vacuum line exterior, control panels), zone 3 (aisles, drains, walls) and zone 4 (outside the plant, including the roof, which is now a documented vector). Sample each zone at a defined frequency, with zone 1 and zone 2 sampled after cleaning and before production.

Indicator choice. Enterobacteriaceae is the right indicator here, because its presence signals moisture and environmental contamination and it is more likely to be recovered than Salmonella. Track counts as trends, not pass/fail only. Add Salmonella testing on a lower frequency for zone 1 and for vector sites.

Vectors. Swab the specific harborage points this event identified: the bag dump station, the vacuum line joints and interior, filler nozzles, hollow legs, cracked welds, the roof drip path, and any surface that has been wet-cleaned. Re-swab the same sites after corrective cleaning so you can prove the intervention worked.

Triggers. Define numeric action levels in the written program - for example, any zone 1 Enterobacteriaceae above the action level triggers immediate cleaning and sanitizing of the area plus investigation; any zone 1 Salmonella positive triggers a hold review of product made on that line since the last negative, intensified cleaning, and re-sampling before restart. Track consecutive results and require a defined number of negatives before returning to baseline frequency.

Link to product and supplier. Tie environmental results to product codes by date and line, and keep incoming ingredient verification (including the composite sampling that caught this lot) as an independent control, since the no-kill-step process depends on ingredient control for safety.