Illustrative case study. Company names and identifying details are fictional. Technical details reflect real industry practice and current regulation.

Underprocessed canned vegetables — the low-acid cans that didn’t get the full botulinum cook — are the canning industry’s nightmare made routine: the product looks normal, the cans are sealed, and the Clostridium botulinum spores (if present) are viable, waiting for the anaerobic, room-temperature can to become the toxin factory. At Harvest Valley Canning, the underprocessing wasn’t a deviation — it was the scheduled process itself: the process had been “optimized” (shortened) five years earlier for quality (less mushy vegetables, brighter color), the revalidation was skipped (“the change was minor”), and the F₀ — recalculated during the investigation — was 30% below the 12D requirement. The plant had been knowingly (if not understandingly) underprocessing for five years.

Background: a vegetable cannery with an “optimized” process

Harvest Valley Canning (fictional) canned about 120,000 cases of vegetables a month — 95 employees, green beans, carrots, and corn — the retort processes: “established,” the operators said — but the green bean process had been shortened five years ago (from 45 minutes to 38 minutes, the quality optimization), the change approved by the plant manager (for the texture improvement), and the revalidation — the heat penetration study for the shortened process — was “planned” and never performed. The process authority wasn’t consulted (“the change was minor, just seven minutes”). The cans were brighter, firmer — and underprocessed.

Challenge: the incubation failures

The routine incubation testing — the cannery’s verification (sample cans incubated at 35°C, checked for swelling) — started failing: the green bean cans, specifically, showing swollen cans at a rate far above baseline. The microbiology — the swollen cans cultured — found Clostridium spp. (thermophilic spoilage, the underprocessing indicators) — not C. botulinum (the testing didn’t find the pathogen, but the underprocessing meant it couldn’t be ruled out). The incubation — the verification the plant performed — was catching the process failure that the process itself had been hiding for five years.

Investigation: the seven minutes

The thermal recalculation — by the process authority, during the investigation — quantified the shortfall: the 38-minute process (the “optimized” one) delivered an F₀ 30% below the 12D C. botulinum requirement. The seven minutes — “minor,” the plant manager said — were 15% of the process time, but the lethality (the F₀, integrating the time-temperature) fell 30% — the non-linear relationship (the last minutes, at the cold spot’s peak temperature, contribute disproportionately) meant the “minor” shortening was a major lethality reduction. The quality optimization had traded the safety margin for the texture.

The five-year gap — the revalidation never performed — was the systemic failure: the process change (the seven-minute shortening) was a scheduled process change under 21 CFR Part 113 — requiring process authority evaluation, heat penetration validation, and refiled scheduled process. None of it happened: the plant manager approved the change (for quality), the operators implemented it (the retort timers adjusted), and the regulatory requirements were unknown or ignored. The “minor change” bypassed the entire low-acid canned food regulatory framework.

The incubation testing — the hero of the story — was the only control that caught it: the routine verification (often seen as a formality) detected the swollen cans, triggered the investigation, and exposed the five-year underprocessing. The plant’s other verifications — the retort charts (recording the 38 minutes faithfully), the seam checks (the cans sealed perfectly) — all passed, because they verified the executed process, not the adequate process. Only the incubation tested the product’s actual sterility.

Root cause: the unvalidated optimization

1. Process shortened without revalidation. The seven-minute “optimization” — for quality — reduced the F₀ 30% below the 12D requirement. The change was a scheduled process modification requiring authority evaluation; it got a plant manager’s approval instead.

2. Regulatory requirements bypassed. The 21 CFR Part 113 change control — process authority, heat penetration, refiled process — was unknown or ignored. The “minor” change evaded the framework designed to prevent exactly this.

3. Verifications checking execution, not adequacy. The charts, the seams, the routine checks all verified that the 38-minute process was executed — none verified it was adequate. Only the incubation tested the product’s sterility.

Corrective actions: restoring the 12D

Immediate: the green bean retort was shut down — the 38-minute process suspended, the 45-minute (validated) process restored. The five years of product — the scope defined by the process change records — was evaluated: the incubation testing was expanded (every lot, the full five-year window — a massive testing program), and the lots showing any incubation failure were destroyed. The risk assessment (the process authority, the actual F₀ data) concluded the botulism risk was low (the 30% shortfall, while non-compliant, likely still achieved commercial sterility for the actual spore loads) — but the regulatory violation was absolute, and the FDA notification and market withdrawal were completed.

Within 30 days, the process change control was rebuilt: any retort process modification — time, temperature, product, container — now requires process authority evaluation before implementation (the authority is on the approval routing, not consulted after), heat penetration validation for the modified process, and the refiled scheduled process. The plant manager’s unilateral authority over thermal processes was revoked — the quality optimization that caused this can’t recur, because the thermal process isn’t the plant manager’s to optimize.

The incubation testing was strengthened: from routine verification to the program’s cornerstone — increased sampling, the results trended (not just pass/fail), and any incubation failure now triggers a full thermal investigation (not just lot rejection). The other products (carrots, corn — the processes not “optimized”) were revalidated anyway — the five-year gap raised the question, and the validation answered it (they were adequate, the data proved it). Within 90 days, the company’s other canneries were audited for unvalidated process changes, finding one more “optimization” (a temperature reduction, similarly unvalidated) at another facility.

Results: the processes that are proven

Twelve months later: zero incubation failures, all retort processes validated (the current validations, not the memories), and the process change control — the authority on the routing, the validation before implementation — preventing two subsequent “optimization” proposals (both withdrawn when the validation requirement was explained). The green beans are slightly softer than the “optimized” ones; they’re also safe, which the optimized ones weren’t.

Lessons learned: what you’d do Monday morning

Audit your retort processes for unvalidated changes — compare the current operating parameters to the validated scheduled processes. If the times are shorter, the temperatures lower, or the products different from the validation, you have Harvest Valley’s problem: the executed process isn’t the proven one. Restore the validated process or revalidate the change — before the incubation catches it.

Then rebuild your process change control: the process authority evaluates before implementation, not after — and the plant manager doesn’t unilaterally optimize thermal processes. The seven minutes for texture isn’t worth five years of underprocessing. And strengthen your incubation testing: it’s the only verification that tests the product’s sterility — trend it, respect it, and investigate its failures fully.

Yeah, but actually — the “minor” seven minutes is the lesson of this case: 15% of the time, 30% of the lethality — the non-linear math that the plant manager didn’t know and the process authority wasn’t asked about. Your thermal process changes aren’t minor until the authority says they’re minor — and the authority says it after the heat penetration study, not before. The quality optimization (brighter beans, firmer texture) was real; the safety cost (30% of the botulinum lethality) was hidden until the incubation found it. Validate the change, respect the 12D, and remember: the cans look perfect when they’re underprocessed. Only the incubation — and the authority — know the truth.