Allergen cleaning validation is the evidence that a changeover cleaning procedure actually removes an allergen: run the allergen product, clean exactly as written, then test the equipment and the first product made after it, and do it successfully several times in a row. FDA's preventive controls rule doesn't require validating allergen controls, but it does require verifying that they're carried out consistently and work. Buyers and food safety audits often go further and ask for the validation study. Validation is done once, and again when the procedure, the equipment or the products change; verification is the routine afterward, a visual check at each changeover and allergen swabs on a schedule. In the example below, a nut butter maker validates its peanut-to-almond changeover and fails the first trial at one valve. Below are the rules, how to run the trials, the example, and a free trial log.
The short version
- Validation shows a cleaning procedure removes the allergen; verification shows it keeps being done and working.
- FDA's rule doesn't require validating allergen controls; it does require verification.
- Buyers and audits often require validation anyway.
- Run three trials in a row with the same written procedure: swab set sites, test the first product after the clean.
- Use allergen-specific tests; an ATP swab shows residue, not a particular allergen.
- Revalidate when the procedure, equipment or products change.
- TaroStack keeps changeover checks and swab results on each batch record. More below.
The rules
| Rule | What it says | Source |
|---|---|---|
| Allergen controls | Procedures that protect food from allergen cross-contact | 21 CFR 117.135(c)(2) |
| Validation | Not required for food allergen controls or sanitation controls | 21 CFR 117.160(c) |
| Verification | Required: show the controls are consistently implemented and effective, with record review | 21 CFR 117.165(a) |
So the legal floor is verification. A validation study is what a buyer means when it asks "how do you know your clean works?", and it's the best answer to that question even when no rule asks for it. This is general information, not legal advice.
How to run the trials
| Step | What to do |
|---|---|
| Write the procedure | Every step of the changeover clean, in order: disassembly, scraping, washing or dry cleaning, drying, inspection |
| Pick the sites | The hardest places to clean: valves, gaskets, augers, nozzles, corners |
| Run the allergen product | A normal run, so the equipment is as dirty as it gets |
| Clean exactly as written | The usual people, the usual time, no extra effort |
| Swab each site | An allergen-specific test for that allergen |
| Test the first product after | Send the first units made after the clean to a lab for the allergen, against your limit |
| Repeat | Three passed trials in a row with the same procedure |
If a trial fails, change the procedure, give it a new version, and start the count again.
An example: peanut to almond butter
Red Oak Nut Butters, a made-up nut butter maker in Little Rock, Arkansas, I use in these examples, fills peanut butter and almond butter on the same filler, and an almond butter buyer asked for proof that its almond butter is free of peanut. Red Oak sets its limit for the first almond butter jars at 2.5 ppm peanut:
| Trial | Procedure | Hopper | Auger | Filler valve | Nozzle | Conveyor | First jars | Passed? |
|---|---|---|---|---|---|---|---|---|
| Sep 14 | v1 | Negative | Negative | Positive | Negative | Negative | 4.1 ppm | No |
| Sep 21 | v2 | Negative | Negative | Negative | Negative | Negative | Not detected | Yes |
| Sep 28 | v2 | Negative | Negative | Negative | Negative | Negative | Not detected | Yes |
| Oct 5 | v2 | Negative | Negative | Negative | Negative | Negative | Not detected | Yes |
Version 1 didn't take the filler valve apart, and the swab there found peanut; the first jars tested 4.1 ppm. Version 2 takes the valve apart and scrubs it, and passed three times in a row: the clean is validated. From then on, every changeover gets a visual check that the filler is clean and dry, and the filler valve is swabbed once a month. Red Oak's whole HACCP plan, with labeling as a critical control point, is in its food safety plan; how the label declares peanuts and almonds is in allergen matrix template.
Validation and verification side by side
| Validation | Verification | |
|---|---|---|
| Question | Does this cleaning procedure remove the allergen? | Is it being done, and does it keep working? |
| When | Once, and again after a change | Every changeover, and on a schedule |
| How | Trials with swabs and product tests | Visual checks, periodic swabs, record review |
| Required by FDA's rule | No | Yes |
How a cleaning schedule and its records are kept is in cleaning log template, and the food safety plan these controls sit in is in what is a food safety plan.
Get the trial log
Download the allergen cleaning validation log. Two sheets: validation trials, each with the procedure version, an allergen swab result at each site, and the first product's test against your limit, where whether each trial passed and whether three in a row passed fill in; and a verification log for the checks after. It's an ordinary Excel file with no macros, and it works in Google Sheets, LibreOffice and Numbers.
Where this stops working
The study proves the clean works when it's done as written. After that, the question is whether it was done on a given day, by whom, before which batch, and a monthly swab on a log sheet doesn't say which batches of almond butter ran after a changeover that was rushed.
How TaroStack does it
TaroStack keeps changeover checks and swab results on each batch record.
Checks on every batch. Fields of your own on the batch record, tick boxes, choices and initials, record the changeover procedure version, the visual check and any swab result for every batch of almond butter.
Lots behind each batch. Each batch records the lots of almonds, salt and jars it consumed, and gets a lot of its own.
If a check fails. A lot on hold is kept out of what can be sold, and each order records the lots each customer got, so one recall report shows who got a batch made after a failed check.
| The job | Log sheet | In TaroStack |
|---|---|---|
| Changeover check before each batch | A separate log | A field on the batch record |
| A swab result | A lab sheet | A field on the batch record |
| Holding a batch | A tag | A hold on the lot |
| Who got a batch | Search the invoices | One recall report |
Batches, batch records with fields of your own, holds and recalls are on every plan, from $49 a month. The first 30 days are free, with no credit card. If you'd like a hand setting up, ask, and we'll do it with you.