At the border: EU RASFF rejections, FDA OASIS refusals, and China GACC sampling
A bag of imported dog food does not simply appear on a shelf. Before it does,
it has to clear a frontier โ and in three of the world's largest pet-food
markets, that frontier is run by three very different machines. The European
Union runs a mandatory cross-border alert network that logs *border rejection
notifications*. The United States runs an import-refusal system (OASIS) behind
which sits a voluntary recall pipeline. China runs customs (GACC) inspection at
import plus a national feed-quality sampling program. This article walks the
three machines, using only what is currently on disk in the data station, and
flags where the frontier data is still being assembled.
EU โ RASFF border rejections (46 records, 2022 peak 13)
Within the RASFF pet-food corpus, filtering for
notificationClassification == "border rejection notification" gives **46 of
255 records (18.0%)** for pet food 2019โ2026. Border rejections are the only
classification that can carry the "no risk" riskDecision โ 7 of 46 do so,
used when the product was rejected for non-safety reasons such as missing
documentation or unauthorized composition.
The hazard distribution inside those 46 border rejections (subject-substring
counts) is tightly led by microbiology and paperwork:
- Salmonella: 11 (23.9%)
- Enterobacteriaceae: 10 (21.7%)
- Additive (unauthorized additive): 10 (21.7%)
This is not a three-way tie at 11/11/11: Salmonella leads at 11 while
Enterobacteriaceae and Additive are tied at 10 each. The top-3 hazards account
for 31 of 46 records (67.4%).
Where are the interceptions happening? The top notifying countries are **Spain
(ES) 12, Italy (IT) 10, Germany (DE) 9** โ EU Mediterranean border posts
intercept the most pet-food shipments. The top origin countries are **China (CN)
11, Turkey (TR) 11, India (IN) 7** โ high-volume third-country origins where
border inspection density is highest. The risk-decision split is: not-serious
26, serious 9, no-risk 7, undecided 4. By year, **2022 saw the peak at 13
border rejections**; 2020 and 2026 each sit at 3 (the latter a partial year).
Caliber note. Source:
rasff-petfood-20260719.json. The topic carries
needs_review: true, and the hazard figures above are subject-substringinference (counts derived by matching the notification subject line). The
snapshot is dated 2026-07-19 and is treated as a frozen extract โ no
later increment is claimed. This is an inferred, review-flagged count, not a
hand-audited tally.
For context against the wider EU pipeline: RASFF totals **255 records
(2019-H1 2026), with the border-rejection hazard peaking in 2022 (13
cases)**, while Salmonella overall peaked in 2023 (21 cases โ 58% of
notifications that year). The EU is the highest-volume of the three systems,
reflecting mandatory cross-border notification plus pan-EU coordination.
US โ FDA OASIS refusals (pending / unverified)
A precise 2026 US border-refusal count cannot be given here, and it would be
wrong to invent one. The dedicated US FDA OASIS refusal topics
(fda_oasis_refusals_microbiological_adulteration_salmonella_listeria_2026 and
fda_oasis_refusals_chemical_contaminants_mycotoxins_histamine_2026) are **not
yet present in the repository**; the structured US refusal dataset is being
built in batch D1. Any specific 2026 US refusal figure is therefore marked
pending / unverified and is not stated.
What the data station can say about the US frontier posture comes from the
recall side of the same pipeline. The US FDA trigger is a **firm-initiated
voluntary recall**; FDA may issue Advisory / Alert / Caution, and Warning
Letters are pre-recall regulatory enforcement. Volume runs at **~3โ5 advisory
events per year for pet food (2018โ2025 average)**. The recurring US hazard
pattern is microbiological and chemical:
- Salmonella dominates raw/frozen pet food (Darwin's, Aunt Jeni's, Texas
Tripe, Performance Dog)
- Listeria monocytogenes co-occurs with Salmonella in raw products
- Aflatoxin in corn-based dry kibble (Sportmix/Mid America Pet Food
2020โ2023 cluster)
- Foreign material (Fromm BeefiBowls 2025)
Severity is classed Class I (fatal/serious) / Class II (remote) / Class III
(unlikely). These are recall-side facts, not OASIS import-refusal lines โ the
border-refusal dimension stays pending until batch D1 lands.
China โ GACC sampling and the 2025 feed-quality plan
China turns pet food away at two linked doors. The first is **Customs (GACC)
inspection at import**: monthly published lists of non-compliant food including
pet food, with voluntary firm recall rare in the CN market (where
consumer-protection enforcement is less mature). Quantified 2025-H1 as **2,093
batches of non-compliant food (all categories, pet food estimated ~5โ10% based
on pet food share of imported food)**; the zhihu 2025-H1 summary cited
"ๅๆฏๅข้ฟ ~12% vs 2024-H1" of 1,871 batches. The CN hazard pattern is
microbiological, documentary, and additive-driven:
- Microbiological (Salmonella, Enterobacteriaceae)
- Adulteration (unapproved animal origin โ e.g., ruminant protein in dog
food forbidden post-BSE)
- Labeling (no Chinese label, missing "for pets only" declaration, missing
GB standard compliance)
- Additive violations (unapproved colorant, preservative)
Common GACC reason categories: ๆชๆฃๅบ (not detected โ passed), ่ดง่ฏไธ็ฌฆ
(product-certificate mismatch), ๆ ็ญพไธๅๆ ผ (label non-compliant), ๆฃๅบๆช็ปๆนๅ็ๅจ็ฉๆบๆงๆๅ
(unapproved animal-derived ingredient detected), ๅฎๅ จๅซ็้กน็ฎไธๅๆ ผ
(safety/hygiene item non-compliant).
The second door is the domestic sampling regime. The **2025 Feed-Quality Safety
Supervision Work Plan (ๅไธๅๆ้จๅๅ ฌๅ , ๅๅ็งใ2025ใ2ๅท)** sets the national
feed (incl. pet feed) supervisory sampling/monitoring program and the disposal
chain for non-conforming products: seal the same batch, suspend production,
re-test, and investigate/transfer to police if still non-conforming. Part I
establishes the feed-quality-safety supervisory sampling (็็ฃๆฝๆฅ) program,
managed by provincial animal-husbandry/veterinary authorities; provincial
supervisory sampling batches must not be fewer than the quota in Annex 1. On
non-conforming products, authorities seal the same batch and suspend production;
if re-test still fails, they investigate and penalise per law, and suspected
crimes are transferred to public security for prosecution. This is a first-hand
Chinese source, C1 substring-verified against the saved official text (pulled
2026-07-29, HTTP 200).
The global recall landscape, 2019โ2026
Step back and the three frontier machines are **three fundamentally different
regulatory systems**, each with a different trigger mechanism (voluntary vs
mandatory), disclosure scope (firm-named vs anonymous), hazard taxonomy (FDA
Class I-III vs RASFF categories vs GACC reasons), and public accessibility
(openFDA / RASFF portal / GACC monthly lists).
Approximate annual volume: US (FDA) โ voluntary + FDA Advisory, 3โ5
advisory/year, firm-named + reason; EU (RASFF) โ mandatory member-state
notification, 30โ45/year (peak 2024: 46), firm-named + reason + distribution;
CN (GACC) โ import inspection refusal, ~150โ300 batches/year (incl. all
food), mostly firm-anonymous. EU RASFF has the highest event volume, reflecting
mandatory cross-border notification plus pan-EU coordination.
The hazard cross-comparison is where the systems rhyme and where they diverge:
Salmonella is Very high in US raw pet food, Very high in the EU, Medium in China;
Listeria is High in US raw, Medium in the EU, Low in China; Aflatoxin is a
cluster event in the US, Sporadic in the EU, Rare in China; heavy metals are
Rare in the US, Moderate in the EU (imported fish meal), Medium in China
(imports); foreign material is Sporadic across all three; adulteration is Rare
in the US, Rare in the EU, Common in China (imports); labeling is N/A for the
US, Moderate in the EU, Very common in the China market.
Refresh cadence differs too: US FDA weekly (fda.gov page monitor), EU RASFF
daily (RASFF portal API), CN GACC monthly (customs.gov.cn publication
schedule).
Sources
This article is built only from topics currently on disk in the data station:
- RASFF_petfood_border_rejection โ 46 border-rejection records of 255
(18.0%), 2022 peak of 13; caliber needs_review: true, subject-substring
inference, frozen snapshot dated 2026-07-19 (no increment claimed).
- Global_pet_food_recall_landscape_2019_2026 โ the three-system integration
framework: US/EU/CN trigger, disclosure, taxonomy, and volume comparison;
255 RASFF records 2019โ2026; GACC 2025-H1 2,093 non-compliant batches.
- cn_feed_quality_sampling_2025 โ first-hand MOA 2025 Feed-Quality Safety
Supervision Work Plan (ๅๅ็งใ2025ใ2ๅท), C1 substring-verified, pulled
2026-07-29.
The two planned fda_oasis_refusals_* topics were not on disk at writing
time, so the US OASIS border-refusal section is reported as pending/unverified
rather than fabricated.