Homeโ€บArticlesโ€บFrom 255 RASFF alerts, Salmonella still wears the crown

From 255 RASFF alerts, Salmonella still wears the crown

If you wanted a single number to sum up what goes wrong in pet food

across the EU and its trading partners, it would be this: of **255

pet-food notifications** logged in the EU's Rapid Alert System for Food

and Feed (RASFF) between June 2019 and July 2026, 99 โ€” 38.8% โ€”

mention Salmonella in the subject line. No other hazard comes close.

This article is built entirely from the project's RASFF pet-food

snapshot (rasff-petfood-20260719.json, 255 records, extracted

2026-07-19). A calibration note up front, because the data demands it:

the dataset is a **point-in-time snapshot with no incremental updates

since 2026-07-19**, and several of the underlying topic files carry

needs_review: true. The RASFF feed provides **no structured hazards

array** โ€” every hazard figure below is derived by case-insensitive

substring matching on the free-text subject field, so categories are

broad and intentionally non-exclusive (a single notification can land in

several). Treat the percentages as well-grounded estimates from one

snapshot, not a live dashboard.

Salmonella: still the dominant hazard

Salmonella is "by far the leading hazard in RASFF pet food

notifications: 99 of 255 records (38.8%) mention Salmonella in the

subject field." The runner-up is Enterobacteriaceae at 37 records

(14.5%), then "additive" (unauthorized additives or composition issues)

at 20 records (7.8%). Salmonella shows up in notifications across every

classification and every year since 2019.

Its single-year peak was 21 cases in 2023, out of that year's 36

total โ€” a 58.3% share, the highest single-year rate in the dataset. The

year-by-year Salmonella burden runs: 2019 = 3, 2020 = 10, 2021 = 13,

2022 = 16, 2023 = 21, 2024 = 16, 2025 = 16, 2026 = 4 (year to date).

The 2023 spike is "mostly driven by raw pet food and dog chew

contamination."

The seven-year volume curve

Total pet-food notifications per calendar year (year taken from

ecValidationDate):

  • 2019: 5
  • 2020: 27
  • 2021: 31
  • 2022: 45
  • 2023: 36
  • 2024: 46 โ† peak
  • 2025: 40
  • 2026: 25 (year-to-date, partial coverage through 2026-07)

The 2020 jump from 5 โ†’ 27 "coincides with the pet adoption surge and EU

regulatory tightening of microbiological criteria under Reg 2019/626,"

which entered application on 2021-03-21. The 2024 peak of 46 reflects

"increased post-COVID surveillance plus expanded categories tracked by

RASFF." The 2026 partial count (25 through July) "projects to ~40โ€“45 if

the second half matches typical yearly patterns."

What else shows up

Beyond Salmonella, the hazard structure for the 255 records breaks down

as:

  • Microbiological โ€” Salmonella: 99 (38.8%)
  • Microbiological โ€” Enterobacteriaceae: 37 (14.5%)
  • Physical โ€” all combined (plastic, glass, metal, foreign body): 19

(7.5%) โ€” of which plastic 8, foreign body 5, metal 4, glass 2

  • Chemical โ€” heavy metals (lead, cadmium, mercury): 4 (1.6%)
  • Chemical โ€” dioxin/PCB: 2 (0.8%)
  • Chemical โ€” vitamins/minerals: 3 (1.2%)
  • Unclassified (no hazard keyword matched): 95 (37.3%)

Microbiological hazards overall account for 136 of 255 (53.3%). The

large "Unclassified" group (95) is an honesty check, not a gap to paper

over: its subjects include "unauthorised substance," "additive,"

"non-compliance," and packaging defects that are reportable but not

hazard-descriptive.

How serious, and howๅ…ฌๅผ€

RASFF splits notifications into four classifications: **Information

notification for follow-up 113 (44.3%), Alert notification 68

(26.7%), Border rejection notification 46 (18.0%)**, and

Information notification for attention 28 (11.0%). Alerts are where

the teeth are โ€” of the 68 alerts, 54 (79.4%) are "serious." Border

rejections are the only class carrying "no risk" findings (7 of 46).

Severity over time (serious + potentially serious share) peaked at

47.2% in 2023, the year RASFF introduced the "potentially serious"

and "potential risk" tiers and the "undecided" backlog dropped from 6

(2022) to 1 (2023) โ€” a sign the system tightened its call. One more

transparency point: only 19 of 255 (7.5%) records are

published=true on the public portal; the rest are visible only via the

backend search API.

The regulatory floor outside the EU

The numbers explain why jurisdictions set a hard line on Salmonella. In

China, requirements sit in the hygiene tables of **GB/T 31216-2014 (dog)

and GB/T 31217-2014 (cat): Salmonella must be "not found" in a 25 g

sample**, tested per GB/T 13091. Japan's MAFF has gone public with an

alert to manufacturers and sellers: Salmonella spp. "is being detected

in pet food distributed in Japan (domestic, jerky etc.)," it can "cause

diarrhea in dogs and cats," and is "very dangerous" for puppies, kittens

and weakened animals โ€” pointing at under-cleaned equipment, insufficient

heat-drying, and unhygienic post-drying handling as the likely causes.

Why this matters

For a brand shipping across borders, the lesson is blunt: Salmonella is

not a one-year blip but a structural, seven-year constant at the top of

the hazard list, and raw pet food plus dog chews are its favorite

vehicles. The 2024 volume peak and the 2023 severity peak suggest

regulators are looking harder, not relaxing. The 37.3% unclassified share

is a reminder that the public picture of pet-food risk is itself

incomplete โ€” exactly the kind of gap a monitoring upgrade (see the

project's D5 RASFF incremental-monitoring batch) is meant to close.


*Sources: project RASFF topics RASFF_petfood_salmonella,

RASFF_petfood_year_by_year, RASFF_petfood_hazard_categories_by_subject_keyword,

RASFF_petfood_classification_decoded, RASFF_petfood_risk_severity_by_year

(snapshot rasff-petfood-20260719.json, 255 records, 2026-07-19; method

per Reg 2019/626, eur-lex CELEX:32019R0626); CN_salmonella_zero_tolerance

(GB/T 31216/31217-2014); jp-microbial_004 (Japan MAFF pet-food Salmonella

alert). Hazard figures are subject-substring inferences on a point-in-time

snapshot with no incremental updates; several source topics carry

needs_review: true.*