Allergen Recalls & CAPA: Prove the Corrective Action Worked

TL;DR

Undeclared allergens were the leading cause of U.S. food recalls for the third straight year: 124 of 320 FDA and USDA recalls in 2025 (39%). The sesame-driven spike of 2023 has faded; the allergen problem hasn’t.
Repeat allergen failures usually trace to verification. A closed CAPA documents the fix, but nothing confirms it held or warns you when the same conditions return.
Four failure points keep resurfacing: changeover label checks, undocumented rework, cleaning verification at changeover, and supplier changes that never reach labeling. FSIS names the same causes in its own allergen verification directive.
FDA-regulated plants must verify that allergen preventive controls are consistently implemented and effective (21 CFR §117.165), with PCQI record review within 7 working days. FSIS-regulated plants carry the same obligation through 9 CFR §417.3 corrective action and §417.4 verification requirements.
Every GFSI scheme now requires proof that a corrective action was effective, not just closed. SQF Edition 10 (audits from January 2027) adds a dedicated change-management clause that ties the two together.
The fix is tying CAPA verification to production records so you see a repeat failure during the shift, not at the next audit.

Six months ago, your quality manager flagged a labeling error at changeover. You filed a CAPA, updated the procedure, and closed it out. Last week the same root cause showed up on a different line, on a different shift.

The corrective action closed, but nothing in your system checked whether it held.

You know allergen control. For most established programs, the gap is verification. A CAPA program documents what went wrong and what you did about it. It rarely captures whether the fix held, or whether the conditions behind the first failure are building again on today's shift.

Undeclared allergens led U.S. food recalls again in 2025

The U.S. PIRG Education Fund’s “Food for Thought 2026” report counted 320 food recalls announced by FDA and USDA in 2025, up from 296 in 2024. Undeclared allergens were linked to 124 of them, or 39%, more than any other cause. Milk was the single most common undeclared allergen.

Looking at the three-year run, the pattern is clear. In 2023, allergens caused 154 of 313 recalls (49%), a 27% jump driven largely by sesame, which the FASTER Act made the ninth major food allergen on January 1, 2023. In 2024, the sesame effect faded: sesame-only recalls dropped from 13 to seven, and allergens fell to 101 of 296 recalls (34%). In 2025, they climbed back to 39% with sesame barely a factor. The spike came and went. The underlying failure rate didn’t move.

. Sesame was a one-time change-management stress test, and the industry mostly passed it within a year. The failures that remain are the ordinary ones: a label roll, a rework decision, a supplier spec. PIRG counted at least 31 recalls in 2025 that cascaded from an ingredient another company had already recalled. A CAPA system built around reacting to incidents isn’t designed to catch any of that.

The pattern holds on the FSIS side. Of the 42 meat, poultry and egg product recalls FSIS logged in 2025, undeclared allergens were the second-leading cause at nine, behind foreign material, plus seven of the year’s 24 public health alerts. FSIS’s own analysis of those events found that most trace to a change in ingredient supplier, product in the wrong package, a misprinted label, or a formulation change.

Four root causes that keep resurfacing

U.S. regulators don’t publish root-cause breakdowns for allergen recalls. Food Standards Australia New Zealand (FSANZ) does. Across 197 undeclared allergen recalls from 2021 to 2025, packaging and labeling errors were the leading root cause, followed by accidental cross-contact and supplier verification failures. In FSANZ’s last fully tabulated dataset (2017–2022), packaging errors made up about 48% of identified root causes and supplier verification about 22%. This is Australian data, so treat it as directional for U.S. plants. Every one of those causes runs through a control your CAPA program should be verifying.

Here is where those controls tend to fail in practice.

Changeover label verification

When a line moves between SKUs with different allergen profiles, mislabeling concentrates. A startup check won't catch an error that creeps in mid-run, like the wrong label roll loaded after a break. If the CAPA from your last labeling incident lives in a spreadsheet and the changeover procedure lives in a binder, nothing confirms the fix made it into how the line actually runs.

Undocumented rework

Rework decisions get made on the floor under production pressure. When allergen-containing rework goes into a batch of an allergen-free product with no record and no label review, cross-contact risk builds. A CAPA that tightens rework segregation can't be verified as effective if the rework log never reaches your quality system before the product ships.

Cleaning verification gaps at changeover

For FDA-regulated plants, the Preventive Controls rule requires written allergen controls under 21 CFR §117.135(c)(2), covering both cross-contact and labeling. Section 117.165 adds verification: you have to confirm your preventive controls are consistently implemented and effective, and a preventive controls qualified individual (PCQI) must review, or oversee review of, monitoring and corrective action records within 7 working days of creation, unless a longer period is justified in writing. For FSIS-regulated plants, 9 CFR §417.3 requires that corrective action identify and eliminate the cause, restore control, and establish measures to prevent recurrence, while §417.4 requires ongoing verification and §417.5(c) requires pre-shipment record review. Under either regime, a documented procedure doesn’t meet the requirement on its own..

The failure pattern is familiar: a cleaning procedure passes verification for several changeovers, then fails on a rushed overnight run after someone reconfigured a piece of equipment. If your verification is periodic instead of tied to each changeover, the failure may not surface until a customer complaint or a recall.

Supplier changes without CAPA review

A seasoning blend that was peanut-free last quarter gets a new component. Or your supplier switches its own supplier. Or an ingredient you received last month is recalled upstream. The notice reaches purchasing but not your regulatory document control, packaging, or allergen labeling teams, so the label keeps declaring the product free of an allergen that’s now in it. FSIS inspectors are specifically directed to scrutinize an establishment’s formulation and labeling after a supplier change for exactly this reason. If your allergen matrix, supplier specs, and label approval workflow aren’t connected, nothing flags that a supplier change needs a CAPA-level review of the controls that depend on it. See how supplier compliance management fits into this loop.

Why paper-based CAPA programs miss the repeat failures

Each of those root causes probably has a documented corrective action in your files: an updated SOP, retrained operators, an added supplier review step. The failures still came back.

Blue Bell Creameries described what paper did to its quality data before it moved to digital records: "We had a lot of binders, filing cabinets. We tried Excel. It's just time consuming and not user-friendly… things are scratched out, scribbled on, wrong color ink." The data was inconsistent and couldn't support real-time decisions across departments or facilities.

A CAPA in a binder records what happened and what you did. It can't show whether the fix held, and it can't warn you when the same conditions are forming on the current shift. The regulations draw the same line. Under 21 CFR §117.150, corrective action procedures must include steps to reduce the likelihood that the problem recurs, and §117.165 requires verification that the controls are consistently implemented and effective. Under 9 CFR §417.3(a), corrective action must include measures to prevent recurrence, and an unforeseen deviation triggers reassessment of the HACCP plan under §417.3(b). A documented corrective action is not a verified root-cause closure, under either agency.

Recalls are expensive. The most widely cited benchmark, a 2011 Grocery Manufacturers Association study (GMA is now the Consumer Brands Association), put the average direct cost of a recall to a food company at $10 million, before brand damage and lost sales, and found that more than half of companies surveyed reported a total impact above that figure. That number is 15 years old; nothing about the cost of a recall has gotten cheaper since. As one customer told us, most teams “feel like they’re fine until they’re not.” In a repeat failure, the system didn’t catch the problem in real time, and the first CAPA didn’t prevent the second event.

What changes when CAPAs connect to floor data

Consider what shifts when your CAPA management process draws on production records as they're created instead of end-of-shift paperwork.

The quality manager at Egglife, whose product is built on eggs as the primary ingredient, described the shift: "I started my career in pen and paper… not having to wait till the end of the day when you're doing paperwork review to realize that you had a failure on the line is huge". Catching a failure during the shift changes your options. The batch is still on the line, containment is possible, and you open the CAPA against a complete record of what happened instead of a reconstruction.

When CAPAs open directly from production records, such as a failed in-process allergen check, an incomplete changeover form, or a rework entry outside defined limits, the chain of evidence from failure to corrective action carries through without manual re-entry. Your supervisor can see during the shift whether the verification step from your last allergen CAPA was completed, without waiting for the next audit.

For supplier-driven allergen changes, a structured supplier compliance workflow ties specification updates to the controls that depend on them, so a change can’t reach purchasing without also reaching quality. Supplier corrective action requests (SCARs) carry the same closed loop upstream when the nonconformance starts outside your facility. This is also what SQF Edition 10’s new change-management clause asks for: a documented evaluation, before the change goes live, of whether it affects allergen status, a CCP or preventive control, cleaning, or validation.

What your certification scheme already requires

Effectiveness verification isn’t a software concept. Every GFSI-benchmarked scheme writes it into the corrective action clause, and allergen management is a dedicated requirement in all four. Here is where to look in the edition you’ll be audited against:

Scheme	Edition audited now	Corrective action / effectiveness	Allergen management	What’s next
SQF Food Safety Code	Edition 9	2.5.3 Corrective and Preventive Action	2.8.1 Allergen Management	Edition 10 published March 2026; audits from Jan 2, 2027. Adds Change Management (2.3.5) and tighter CAPA effectiveness verification
BRCGS Food Safety	Issue 9	3.7 Corrective and preventive actions	5.3 Management of allergens	Issue 10 expected 2H 2027
FSSC 22000	Version 6 (V7 upgrade audits May 2027–Apr 2028)	ISO 22000:2018 cl. 8.9.3 and 10.1	Additional requirement 2.5.6	Version 7 published May 2026; V6 audits end April 30, 2027
IFS Food	Version 8	5.11 Corrective actions	4.20 Allergen risk mitigation	No transition pending

The language differs, the expectation doesn’t. BRCGS asks for verification that the corrective action was effectively implemented. ISO 22000 asks you to review the effectiveness of the corrective action taken. SQF Edition 10 goes furthest, calling for a single consolidated CAPA program with documented root cause, verification of effectiveness, and recurrence trending. A repeat finding on the same root cause is hard to defend with a closed CAPA under any of them, and under Edition 10’s Core Clause scoring it costs more points than it used to.

Edition 10’s audit date is January 2027. The reason to act now isn’t the calendar. It’s that the four failure points above don’t wait for a transition audit, and the FSMA and FSIS obligations described earlier are already in force. When verification happens in the system as the work gets done, the audit trail is a byproduct of the work, not a documentation project before every audit, whichever edition the auditor brings.

Three questions to test your current program

Run these checks against your program now, before the next recall runs them for you.

Can a supervisor on the floor see right now whether the corrective action from your last allergen CAPA has been verified as effective? If answering means pulling a file or calling someone, your verification gap is open.
If a supplier changed the allergen status of an ingredient you use, would your CAPA process flag the dependent controls, such as labels, scheduling, and changeover procedures, for review? If that depends on someone remembering to check, your supplier change gap is open.
If the root cause behind an allergen CAPA from twelve months ago is developing on the floor today, would anyone know before the next batch ships? If the only detection is a scheduled audit or a customer complaint, your prevention gap is open.

If any of those answers made you uncomfortable, a CAPA and allergen control gap assessment is a faster starting point than a full system evaluation. You can also ask for a walkthrough built around your audit schedule and facility type: see what that looks like for your operation.

Frequently asked questions

Most CAPA programs record the failure and the corrective action but don't verify that the fix held. If the changeover procedure, rework log, and supplier specs aren't connected to the CAPA, the conditions that caused the first failure can return on another line or shift without anyone seeing it. Closing a CAPA documents the response. It doesn't prove the root cause is gone.

They are the leading cause every year. The U.S. PIRG Education Fund counted 320 food recalls announced by FDA and USDA in 2025, and undeclared allergens were linked to 124 of them (39%). The share was 34% in 2024 and 49% in 2023, when sesame’s addition as the ninth major allergen drove a one-year spike. On the FSIS side, undeclared allergens caused nine of 42 recalls and seven of 24 public health alerts in 2025.

Under 21 CFR §117.135(c)(2), allergen controls must be written and cover both cross-contact and labeling. Under §117.165, you must verify that preventive controls are consistently implemented and effective, and a preventive controls qualified individual must review (or oversee review of) monitoring and corrective action records within 7 working days of creation, unless a longer period is justified in writing. Under §117.150, corrective actions must include steps to reduce the likelihood the problem recurs.

FSIS-inspected establishments aren’t covered by 21 CFR Part 117. Allergens are addressed through the hazard analysis under 9 CFR Part 417, and FSIS’s compliance guideline recommends controlling them through the HACCP plan, Sanitation SOPs, or prerequisite programs. Corrective action under §417.3 must identify and eliminate the cause and prevent recurrence; §417.4 requires ongoing verification, and §417.5(c) requires pre-shipment review of records. [10] Under FSIS Directive 7230.1, inspection personnel run a recurring “Big 9” formulation and labeling verification task, scheduled more often after a supplier change, a label change, or an allergen-related recall or complaint.

FSMA doesn’t. 21 CFR §117.160(c) says you do not need to validate the food allergen controls in §117.135(c)(2), though verification is still required under §117.165. Your certification scheme is a different matter: BRCGS Issue 9, SQF and FSSC 22000 all expect allergen cleaning procedures to be validated as effective and then verified at a risk-based frequency. If you’re GFSI-certified, plan to the scheme’s bar, not FSMA’s floor.

U.S. agencies don’t publish root-cause breakdowns, but Food Standards Australia New Zealand does. Across 197 undeclared allergen recalls from 2021 to 2025, packaging and labeling errors were the top root cause, followed by accidental cross-contact and supplier verification failures. In FSANZ’s last tabulated dataset (2017–2022), packaging errors accounted for about 48% of identified causes and supplier verification about 22%. FSIS’s own directive attributes most of its allergen recalls to supplier changes, wrong packaging, misprinted labels, and formulation changes.

Tie verification to the event that caused the failure. Check the fix at each changeover instead of on a periodic schedule, confirm rework entries and label reviews appear in your quality system before product ships, and review CAPA records against production data. Then check whether the same root cause is developing on the current shift, not just whether the CAPA is closed.

Yes, in two ways. Edition 10 (published March 2026, audits from January 2, 2027) asks for a single consolidated CAPA program with documented root cause analysis, verification that each action was effective, and trending for recurrence. It also adds a dedicated change-management clause (2.3.5) requiring a documented risk evaluation before changes to formulations, ingredients, suppliers, labels, equipment or processes go live. BRCGS, FSSC 22000 and IFS already expect effectiveness verification in their current editions, so the practice isn’t new; Edition 10 makes the gap more expensive on the scorecard

Allergen Recalls and CAPA: How to Prove the Corrective Action Worked