Kim et al. 2018 — Arsenic, cadmium, lead, and mercury in dry dog foods grouped by primary animal ingredient
This study measured total arsenic, cadmium, lead, and total mercury by ICP-MS in 51 over-the-counter dry dog foods (17 fish-based, 17 poultry-based, 17 red-meat-based), reported concentrations as mg/Mcal metabolizable energy, and compared canine intakes to published US human total-diet-study intakes. Fish-based diets had significantly higher arsenic, cadmium, and mercury; red-meat-based diets had the highest lead. The source is routeable to dry-kibble pet-food evidence; it is not routeable to human ingredient or product pages despite the appearance of “fish,” “poultry,” and “red meat” in formulation labels.
Key numbers
51 commercial dry dog foods analyzed: 17 fish-based, 17 poultry-based, 17 red-meat-based. All values reported as mg/Mcal metabolizable energy (mg per 1,000 kcal); the authors converted ICP-MS mg/kg-as-fed output by dividing by manufacturer-reported energy density. Sample comparator: NRC mean US adult intake for a 79 kg male consuming 2,900 kcal/day.
Median (range) mg/Mcal by primary animal ingredient (published Table 1):
- Arsenic (total): fish 0.343 (0.025–1.104); poultry 0.054 (0.007–0.133); red meat 0.037 (0.007–0.134). Fish > poultry, fish > red meat (p < 0.01).
- Cadmium: fish 0.027 (0.014–0.215); poultry 0.015 (0.005–0.063); red meat 0.013 (0.008–0.047). Fish > poultry, fish > red meat (p < 0.01).
- Mercury (total): fish 0.0082 (0.0010–0.0139); poultry 0.0008 (0.0004–0.0023); red meat 0.0012 (0.0004–0.0064). Fish > poultry, fish > red meat (p < 0.01).
- Lead: fish 0.049 (0.018–0.325); poultry 0.037 (0.019–0.305); red meat 0.091 (0.032–1.621). Red meat > poultry (p < 0.03); fish not significantly different from either.
Reported US human comparator intakes (mg/Mcal, Table 1, derived from cited total-diet studies): As 0.017–0.02; Cd 0.0034–0.0138 (mean 0.0086); Hg 0.0016; Pb 0.006.
Fold-difference vs human comparator (authors’ own arithmetic in Results):
- As: fish-based median 16× human daily intake per Mcal; fish-based maximum 55× the canine median per Mcal. No fish-based diet was below the human daily intake. Poultry was below human intake in 3 of 17; red meat below in 5 of 17.
- Cd: fish-based median ~2× the upper bound of human daily intake; fish-based maximum 15× the upper bound. Poultry median slightly above the human upper bound; red meat median within the human range.
- Hg: fish-based median 5× the human daily intake; fish-based maximum >8× the human daily intake. Poultry and red-meat medians close to the human comparator.
- Pb: red-meat median 15× the human daily intake; red-meat maximum 270× the human daily intake. Poultry median 6× human; fish median 8× human.
All 51 samples were above the ICP-MS lower limit of quantification (0.001 mg/kg as fed) for every metal measured.
Methods
Sample handling and digestion: each 100 g bag was pulverized individually with plastic utensils in plastic vessels to avoid cross-contamination, and triplicate 5 g aliquots were submitted to Eurofins Central Analytical Laboratories (New Orleans, LA, USA). Aliquots were digested via microwave-assisted nitric acid procedure and oven-dried at 40 °C following minor modifications of Noel et al. (2005) (J AOAC Int 88:1811).
Instrumentation: inductively coupled plasma mass spectrometry on an Agilent 7500ce (Agilent Technologies, Santa Clara, CA) coupled with a Mira Mist nebulizer (Burgener Research, Mississauga, ON) operated in hydrogen mode. Lower limit of quantification 0.001 mg/kg as fed for each of the four metals. Intra-assay precision (1 sample, 10 replicates) and inter-assay precision (5 samples, 3 separate days) were below 5% and 7% respectively for all four analytes.
Speciation: total arsenic and total mercury only. The authors explicitly note that valency and organic/inorganic forms were not determined and that this is a study limitation (Discussion, “limitations” paragraph). Inorganic As and methylmercury fractions are not reported.
Normalization: mg/kg-as-fed output was divided by manufacturer-reported metabolizable energy density (kcal/kg) to express each metal as mg/1,000 kcal (mg/Mcal). Each product’s energy density was obtained from the product label or via direct manufacturer contact.
Statistics: Shapiro-Wilks rejected normality for all four metal distributions within each ingredient group. Group comparisons used Kruskal-Wallis with Dunn’s post-hoc; descriptive statistics reported as median (range). Software: JMP 12.0 (SAS Institute, Cary, NC). Canine intakes per Mcal were compared to NRC-derived human intakes for a 79 kg male consuming 2,900 kcal/day (NRC Recommended Dietary Allowances 1989).
Implications
Pet-food occurrence evidence and animal-ingredient comparison context for dry kibble in the US market. The strongest finding for downstream wiki use is the per-Mcal magnitude of the metal load across primary animal protein sources: fish-based dry kibble loads more As, Cd, and Hg per unit energy than poultry or red-meat-based kibble, while red-meat-based kibble loads more Pb. The authors attribute the fish signal primarily to use of concentrated fish meal (cited as ~5× higher Hg than fresh fish, with >80% in methylmercury form in fish meal per Johnston & Savage 1991), and the red-meat lead signal to potential lead-shot exposure in game-meat ingredient streams (bison, venison) and to lead-paint-exposed terrestrial sources.
Routing scope: this source is for pet-food product-category pages only. It does not contribute occurrence values to human-food ingredient pages (fish, meat-and-poultry, etc.) even though those words appear in the formulation labels; the analyte was the finished dry kibble, not the underlying ingredient as a human food.
Verification notes
Brand handling: the paper lists 34 manufacturer company names with product counts in the Analyzed Foods section (largest contributors 4 products each). Per Part 12 brand firewall (strict reading locked 2026-05-17), the manufacturer list is not reproduced and brand names are not named here either. Contamination values are reported only at the ingredient-group level (fish/poultry/red-meat), which is the appropriate aggregate.
Speciation: total As and total Hg only; iAs and MeHg fractions not reported. The metals frontmatter uses tAs and tHg accordingly.
Matrix: pet-food is retained as the matrix because it is the controlled-vocabulary slug already used by the pet-food source subcorpus (du2025, kalicharan2025, and others) and is the least misleading choice for dry kibble. Product slug is pet-food-dry-kibble; the alternative pet-food product slugs (canned-wet, freeze-dried, raw, semi-moist) are not routed because the study explicitly restricted to dry foods.
Ingredients frontmatter empty by design: per Part 12 / Part 5b, dog-food protein source labels do not route to human ingredient pages. No pet-food-specific ingredient pages currently exist in the wiki taxonomy.
2026-05-28 (Claude Code, merge-enhance pass on auto-fetched duplicate):
A second copy of this paper arrived in raw/manual-fetch/seasonal-geographic-variance/auto-fetched/ing-meat-poultry-thg_2018_10-3389-fvets-2018-00264.pdf (a third copy as ing-meat-poultry-tas_2018_10-3389-fvets-2018-00264.pdf also sits in the same folder). The paper had been ingested previously against the ing-meat-poultry-pb_2018_… copy in ingested/; raw_path continues to reference that copy. DOI 10.3389/fvets.2018.00264 identity-checked across all three filenames. Page was advisory-flagged in routing_malformed.csv for missing optional routing-input fields (the empty ingredients: [], which is intentional for pet-food sources). This pass enhances rather than rewrites.
Defects corrected from the prior wiki revision (updated 2026-05-27):
- Key numbers expanded from 7 bullets to full Table 1 transcription with all four metals × three ingredient groups × median + range, plus the human-comparator row and the authors’ fold-difference arithmetic from Results (16× / 55× / 15× / 270× and so on). The prior version omitted ranges for poultry/red-meat cadmium and mercury, omitted the lead range maxima entirely, and omitted all fold-over-human numbers.
- Methods expanded from a 2-sentence summary to a full transcription of digestion (microwave-assisted HNO₃, Eurofins New Orleans), instrumentation (Agilent 7500ce with Mira Mist nebulizer in hydrogen mode, LOQ 0.001 mg/kg as fed, intra/inter-assay CV <5%/<7%), statistical method (Shapiro-Wilks → Kruskal-Wallis + Dunn’s post-hoc, JMP 12.0), and the explicit no-speciation caveat. Vendor and software names retained per Part 12 Exception 2 (scientific-method vendor/material names are not brand-firewall violations).
- Implications section rewritten. The prior version contained a single routing-guidance sentence (“It should not be routed to human meat, poultry, or fish ingredient or product pages”). Routing guidance moved into Verification notes; Implications now states the substantive finding (per-Mcal metal load by animal ingredient) and the authors’ mechanistic explanation (fish meal concentration of Hg, lead-shot in game meat). No HMT&C threshold proposals or consumer risk advisories added.
sample_populationexpanded from a one-line description to specify the retailer breakdown (47 online, 1 grocery, 1 pet retailer, 2 Cornell VMC).- Title heading on body changed from “Kim et al. 2018 - Arsenic, cadmium, lead, and mercury in dry dog foods” to add “grouped by primary animal ingredient” — the protein-source comparison is the paper’s primary axis.
- Lead paragraph rewritten to summarize the paper’s scope and the routing scope decision in one paragraph (the prior 2-sentence version was thin).
Preserved unchanged per the v2 skill rules: cite_key, raw_handle: auto-fetched, raw_path, license, doi, source_type, evidence_tier, metals, ingredients: [], products, matrices, jurisdictions, sample_n.
No brand-firewall violations introduced or remaining. No HMT&C drift. Total As / total Hg consistent with metals frontmatter (tAs, tHg).
2026-05-28 (fresh-context audit subagent, verdict REVISE):
Checks 1, 2, 3, 5 returned ✅. Check 4 raised one ⚠️ finding: the Brand-handling paragraph above named two manufacturers as examples (“e.g., Nestle-Purina n=4, Zignature n=4”). Verified against source — the paper does list these on page 2, but per the strict Part 12 reading (locked 2026-05-17), ANY brand name in source-page content must be removed unless it falls under Exception 1 (regulatory-event subject) or Exception 2 (scientific-method vendor/material). These are commercial product manufacturers, not method vendors, so neither exception applies. Corrected — illustrative brand names removed; rephrased to “largest contributors 4 products each”. All other content (numerical fidelity for Table 1, fold-difference arithmetic, methods, speciation, taxonomy slugs, HMT&C firewall) audited clean.
Wiki pages this source may touch
Page history
The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.