Recall or not? Reading the regulatory spectrum from the DCM investigation's close
Part two of the regulatory-spectrum companion to the DCM evidence-chain article.
A spectrum, not a switch
"Recall" is the word most people reach for when a pet-food scare hits the news. But the regulatory record shows a recall is only one point on a wider spectrum of government responses — the same agency can mount a 4.5-year investigation and close it without pulling a single bag from the shelf, yet trigger a Class I withdrawal on a contamination finding. The diet-associated dilated cardiomyopathy (DCM) story and the standing FDA recall framework sit at opposite ends of how enforcement power is actually exercised.
The DCM close: a 4.5-year investigation, no recall
On December 23, 2022, FDA updated its consumer-facing DCM investigation page with what became the most consequential sentence of the entire episode: "FDA does not intend to release further public updates until there is meaningful new scientific information to share." That date is the official investigation closure for the FDA public record.
The reason no recall followed is spelled out on the same page. The agency stated that, "by themselves, they do not supply sufficient data to establish a causal relationship" between the reported diets and DCM — a methodological position, not a denial that diet might matter. As the companion science topic puts it, the closure should be read as "no causal-link statement has been made" rather than "no causal link exists."
The practical result is blunt: "The DCM signal is not a recall event — no product was withdrawn from market based on the FDA investigation." A vast reporting effort — by November 1, 2022 the page carried a final public case count, and the underlying timeline shows >1,100 reported to FDA through 2020-09, with 340 dogs + 9 cats medical records reviewed and 202 dogs + 6 cats confirmed DCM — nonetheless ended with no enforcement withdrawal, because the adverse-event data did not establish causation.
The signal was real enough to act on clinically: in the investigation's diet analysis, 91% of reported products were grain-free and 93% were pea- and/or lentil-rich, and 107 fully or partial recovery cases improved after a diet change. But a recoverable cardiac pattern is not the kind of verifiable hazard the recall statute is built to address.
The recall machinery: three classes
Where the DCM file stayed open-and-unresolved, routine contamination enforcement runs through a fixed, classification-driven ladder:
- Class I (most serious): reasonable probability that use of the product will cause serious adverse health consequences or death. Textbook examples include Salmonella in high-moisture food, aflatoxin >100 ppb, melamine >50 ppm, and vitamin D >10,000 IU/kg in dogs.
- Class II: temporary or medically reversible health consequences, or probability of serious consequences is remote. Examples include low-level aflatoxin 20-100 ppb, vitamin D 5,000-10,000 IU/kg, Listeria monocytogenes in frozen raw food, and low-level heavy metal contamination.
- Class III: violation of FDA labeling or manufacturing regulations where health consequences are unlikely. Examples include net weight discrepancy, mis-ordered ingredient list, and color additive naming error.
Crucially, "FDA may mandate Class I/II recalls or accept manufacturer-initiated recalls; Class III recalls are typically voluntary." The mechanism is triggered by a concrete, measurable hazard, not by a statistical association in voluntarily submitted adverse-event reports.
The contamination lane: Salmonella and friends
This is the lane DCM never entered. When a hazard is discrete and verifiable — Salmonella in high-moisture food under Class I, or Listeria monocytogenes in frozen raw food under Class II — the framework supplies both the trigger and the severity grade. The contrast with DCM is structural: contamination findings give the agency the kind of primary, confirmable fact the classification system is built around, whereas the DCM record, in FDA's own phrasing, "do not supply sufficient data to establish a causal relationship."
A separate topic describing 2026 H1 recall activity lists a Hill's Science Diet product with "potential Salmonella contamination" (Class II) among its examples. That topic is flagged here as pending / unverified: it carries no retrieved date and no source_file on disk (generated from the topic text alone), so its specific figures — "12 Class I/II/III recalls" or "4.2 million lbs" — are not treated as confirmed fact. It is mentioned only to show the contamination-recall shape; the verified recall framework above is the load-bearing source for any concrete claim.
When the agency stays silent: the Lanham Act detour
Because FDA closed without a recall and without a causal finding, the dispute about what grain-free diets "did" migrated into the courts. In February 2024, KetoNatural Pet Foods filed a $2.6 billion Lanham Act false advertising lawsuit against Hill's Pet Nutrition (a Colgate-Palmolive subsidiary), the Morris Animal Foundation, the Mark Morris Institute, and five veterinary researchers — Drs. Lisa Freeman, Joshua Stern, Darcy Adin, Ryan Fries, and John Rush — alleging they had "weaponized" the FDA's DCM reporting system to suppress the grain-free segment.
The two claims were false advertising under the Lanham Act (15 U.S.C. § 1125(a)(1)(B)) and a civil conspiracy under Kansas state law. In November 2025, the United States Court of Appeals for the Tenth Circuit affirmed the dismissal of the complaint. The panel held that the challenged Hill's statements characterized "the FDA investigation record" rather than making actionable establishment claims, and that the veterinarians' statements were "first-party statements by the scientists themselves." The result is a procedural win for Hill's — it adjudicated the lawsuit's pleading, not the underlying science.
That matters for the spectrum argument: when the agency declines to convert a signal into an enforcement action, private litigation becomes the alternative arena — a procedural outcome about commercial speech, not a substitute for the recall authority FDA chose not to exercise on DCM.
Where the authority actually lives: FDA CVM
The body that would have issued any DCM-related recall is the FDA Center for Veterinary Medicine (CVM, established 1984), the primary FDA center responsible for pet food safety oversight. Its jurisdiction covers pet food (dogs, cats, other companion animals), animal drugs, medicated feeds, veterinary devices — with a clear boundary: CVM does NOT have jurisdiction over feed for food-producing animals destined for human consumption.
CVM's Office of Surveillance and Compliance handles four things directly relevant here: pet food adverse event reporting (with over 1,000 formal complaints logged in 2025), inspection of pet food manufacturing facilities, recall authority for non-compliant pet foods, and AAFCO coordination on nutrient profiles. The center operates with 130 staff and approximately USD 25M annual budget for pet food oversight. The adverse-event portal is exactly the channel that fed the DCM case count — and, as the closure statement noted, FDA "is unable to investigate every report to verify or confirm the reported information."
Reading the spectrum
Lined up, the contrast is clean:
- DCM (2014–2022): a large, multi-year signal — >1,100 reported to FDA, 202 dogs + 6 cats confirmed — closed on December 23, 2022 with "no product was withdrawn from market" because the data "do not supply sufficient data to establish a causal relationship."
- Contamination recalls: discrete hazards like Salmonella in high-moisture food or Listeria monocytogenes in frozen raw food trigger a graded Class I/II/III response.
- Private litigation: when the agency stays silent, parties turn to the Lanham Act — a $2.6 billion claim the Tenth Circuit affirmed dismissal of in November 2025, resolving procedure rather than science.
The throughline is that FDA's pet-food power is real but conditional. CVM holds recall authority for non-compliant pet foods, yet on DCM it exercised that authority only to the extent of publishing and retiring a case count. The recall ladder exists for hazards it can measure; the DCM story shows what happens when the hazard is a correlation the agency will not, on the record, call causal.
Sources
- FDA_DCM_investigation_closure_2022_dec — closure date (December 23, 2022), "no causal relationship" statement, "no product was withdrawn from market." (Verified, retrieved 2026-07-20.)
- FDA_DCM_investigation_timeline_2014_2024 — case counts (>1,100 reported; 340 dogs + 9 cats; 202 dogs + 6 cats), diet profile (91% grain-free, 93% pea/lentil), recovery counts. (Extracted via Tavily, 2026-07-20.)
- FDA_DCM_post_closure_science_2023_2025 — "no causal-link statement has been made" vs "no causal link exists"; "strong link … particularly regarding legumes"; Veterinary Sciences 2025;12(11):1106. (10/10 claims verified, 2026-07-21.)
- keto_natural_v_hills_lanham_act_lawsuit_2024_2025 — $2.6B Lanham Act suit (Feb 2024), 15 U.S.C. § 1125(a)(1)(B), Tenth Circuit affirmance (Nov 2025), "procedural win for Hill's." (9/9 claims verified, 2026-07-21.)
- FDA_pet_food_recall_classification — Class I/II/III definitions and examples (Salmonella, Listeria, aflatoxin, vitamin D, labeling defects). (Official web page; C1 source.)
- FDA_CVM_petfood_oversight — CVM established 1984, jurisdiction, 1,000+ complaints in 2025, 130 staff / ~USD 25M budget, recall authority. (Official source; needs_review flag.)
- FDA_pet_food_recall_2026 — PENDING / UNVERIFIED. No retrieved date, no source_file (generated from topic text only). Specific 2026 figures (e.g. "12 recalls", "4.2 million lbs") are NOT treated as confirmed in this article; referenced only to illustrate the contamination-recall shape.