Barber et al. 2025 - Toxic elements in US baby and young-child foods
Barber et al. report an FDA Human Foods Program non-targeted convenience survey of 566 ready-to-eat foods intended for babies, young children, pregnant women, and nursing mothers, collected from the U.S. marketplace in 2023. The study measured total arsenic, cadmium, total mercury, lead, and thallium by ICP-MS, then ran targeted arsenic speciation on rice-containing dry foods/supplements above 50 µg/kg total As and methylmercury analysis on the two fish-containing puree/mixture products. The ingredient-predictor analysis found rice-associated arsenic, root-vegetable-associated lead, and kale/brassica-associated thallium signals, while cadmium and mercury showed weaker or less stable ingredient associations.
Key numbers
Sampling frame and data structure (Methods; Table 1):
- 566 products collected January-December 2023: 165 dry foods/supplements, 394 puree/mixtures, and 7 liquids/drinks.
- Product selection was availability-based and was not market-share-weighted or designed to mimic consumption/sales.
- The 2023 puree/mixture data were combined with FDA’s 2021 puree survey for predictor screening, creating 794 puree/mixture products and 3,714 ingredient-product rows; dry foods/supplements contributed 1,221 rows, and thallium screening used 2,113 rows from 2023 products.
Limits of detection and quantification (Tables 4-6; µg/kg):
| Matrix group | Analyte | LOD | LOQ |
|---|---|---|---|
| Dry foods/supplements (100x dilution) | As | 1.3 | 11 |
| Dry foods/supplements | Cd | 1.4 | 13 |
| Dry foods/supplements | Hg | 0.85 | 7.5 |
| Dry foods/supplements | Pb | 1.4 | 13 |
| Dry foods/supplements | Tl | 2.0 | 18 |
| Puree/mixtures (20x dilution) | As | 0.24 | 2.2 |
| Puree/mixtures | Cd | 0.28 | 2.5 |
| Puree/mixtures | Hg | 0.09 | 0.80 |
| Puree/mixtures | Pb | 0.29 | 2.6 |
| Puree/mixtures | Tl | 0.42 | 3.7 |
| Liquids/drinks (10x dilution) | As | 0.19 | 1.3 |
| Liquids/drinks | Cd | 0.11 | 0.78 |
| Liquids/drinks | Hg | 0.10 | 0.70 |
| Liquids/drinks | Pb | 0.046 | 0.32 |
| Liquids/drinks | Tl | 0.17 | 1.1 |
Occurrence summaries for results above LOD (Tables 7-9; µg/kg):
| Matrix group | Analyte | Maximum | Median | 95th percentile |
|---|---|---|---|---|
| Dry foods/supplements | tAs | 311 | 10 | 95 |
| Dry foods/supplements | Cd | 83 | 12 | 35 |
| Dry foods/supplements | tHg | 3.2* | 1.9 | 2.1 |
| Dry foods/supplements | Pb | 56 | 4.3 | 21 |
| Dry foods/supplements | Tl | 6.6 | 3.0 | 2.5 |
| Puree/mixtures | tAs | 57 | 2.6 | 8.8 |
| Puree/mixtures | Cd | 30 | 2.6 | 13 |
| Puree/mixtures | tHg | 4.0 | 0.14 | 0.21 |
| Puree/mixtures | Pb | 26 | 1.4 | 8.0 |
| Puree/mixtures | Tl | 8.63 | 0.80 | 2.5 |
| Liquids/drinks | tAs | 4.9 | 1.7 | 4.9 |
| Liquids/drinks | Cd | 1.4 | 0.78 | 1.0 |
| Liquids/drinks | tHg | <LOD | <LOD | <LOD |
| Liquids/drinks | Pb | 3.3 | 0.47 | 3.0 |
| Liquids/drinks | Tl | 0.38 | 0.25 | 0.35 |
* Table 7 excludes a total-mercury outlier of 44 µg/kg from the dry-food/supplement tHg summary.
Source-reported whole-survey summaries (Results):
- Total arsenic ranged from <LOD to 311 µg/kg; median 3.28 µg/kg.
- Cadmium ranged from <LOD to 83 µg/kg; median 6.5 µg/kg.
- Lead ranged from <LOD to 56 µg/kg; median 1.7 µg/kg.
- Thallium ranged from <LOD to 8.6 µg/kg; median 0.36 µg/kg.
- Only four products contained total Hg above the LOQ. The highest tHg result was a 44 µg/kg dry food/supplement outlier with cassava flour as the primary ingredient; duplicate analysis confirmed product mean 44 µg/kg with SD 0.5 µg/kg, and methylmercury was not detected above the 4 µg/kg LOD.
Detection/quantification counts and source interpretation (Results):
- tAs above LOQ: 76/165 dry foods/supplements; 225/394 puree/mixtures; 3/7 liquids/drinks.
- All 26 dry foods/supplements with tAs above 50 µg/kg contained rice or rice flour, except one wheat-flour-main-ingredient product.
- Two fish-containing puree/mixtures with wild-caught salmon listed as an ingredient had tAs above 40 µg/kg; arsenate (AsV) was below the 15 µg/kg LOQ in both.
- Cd above LOQ: 65/165 dry foods/supplements; 184/394 puree/mixtures; 1/7 liquids/drinks. The liquid/drink with Cd above LOQ contained pear juice as the main ingredient and measured 1.4 µg/kg Cd.
- Pb above LOQ: 17/165 dry foods/supplements; 96/394 puree/mixtures; 5/7 liquids/drinks.
- The source text reports Tl as “12 of 394 (24%)” puree/mixture products above LOQ, but 12/394 is not 24%; this page preserves the source-reported numerator and denominator and treats the parenthetical percentage as a source-side arithmetic inconsistency.
Predictor-screening and correlation results (Results; Figures 1-5):
| Analyte | Top predictor signal | Model / correlation details |
|---|---|---|
| tAs, dry foods/supplements | Rice | Pearson r = 0.3075, p < 0.0001, 95% CI [0.2559, 0.3575], Cohen’s d = 0.6464; linear regression coefficient = 58.7145, p < 0.0001, R2 = 0.0946. |
| tAs, dry foods/supplements | Apple juice | Pearson r = 0.1402, p < 0.0001, 95% CI [0.0847, 0.1948], Cohen’s d = 0.2831; linear regression coefficient = 42.7159, p < 0.0001, R2 = 0.0197. |
| tAs, puree/mixtures | Rice | Predictor Contribution = 768.33, Portion = 0.1754, Rank = 1; Pearson r = 0.1408, p < 0.0001, 95% CI [0.1091, 0.1722], Cohen’s d = 0.2844. |
| Cd, puree/mixtures | Tomato | Predictor Contribution = 1396.5, Portion = 0.0986, Rank = 1; Pearson r = 0.1574, p < 0.0001, 95% CI [0.1259, 0.1886], Cohen’s d = 0.3188. |
| tHg, puree/mixtures | Salmon ingredient | Predictor Contribution = 5.1982, Portion = 0.4079, Rank = 1; Pearson r = 0.4305, p < 0.0001, 95% CI [0.4040, 0.4564], Cohen’s d = 0.9540. The authors caution that only three puree/mixture products were above Hg LOQ. |
| Pb, puree/mixtures | Sweet potato | Predictor Contribution = 1433.57, Portion = 0.5833, Rank = 1; Pearson r = 0.2316, p < 0.0001, 95% CI [0.2009, 0.2618], Cohen’s d = 0.4761. |
| Tl, puree/mixtures | Kale/brassica signal | Kale: Predictor Contribution = 45.1574, Portion = 0.4439, reported Rank = 3; Pearson r = 0.2107, p < 0.0001, 95% CI [0.1696, 0.2511], Cohen’s d = 0.4310. |
High-percentile ingredient patterns (Results):
- All 9 dry foods/supplements above the 95th percentile for As contained rice or comparable rice ingredients.
- 7 of 9 dry foods/supplements above the 95th percentile for Cd included cereal grains such as oats and rice as a primary ingredient.
- 18 of 20 puree/mixtures above the 95th percentile for Cd contained tomato, spinach, or carrot ingredients.
- 18 of 20 puree/mixtures above the 95th percentile for Pb included sweet potato and/or carrot.
- 9 of 20 puree/mixtures above the 95th percentile for Tl contained brassica vegetables such as kale and/or broccoli.
Methods (brief)
Products were purchased online or in retail locations in the Gainesville, Florida, USA area between January and December 2023. The authors grouped samples as dry foods/supplements, puree/mixtures, and liquids/drinks. Up to three products from multipacks were homogenized; dry foods, ready-to-eat meals, and supplements used IKA grinding mills and disposable grinding chambers, while purees were mixed in 50 mL polypropylene conical tubes. Duplicate analytical portions were taken for all surveyed products to identify potential inhomogeneity or microwave-vessel contamination.
Elemental analysis followed FDA Elemental Analysis Manual (EAM) 4.7: microwave-assisted acid digestion, ICP-MS, and quality controls designed for ultra-low LODs. Analytical portions were 0.5 g for dry foods/supplements, 2.5 g for puree/mixtures, and 5.0 g for liquids/drinks, digested with concentrated HNO3 and H2O2 in PTFE vessels; HCl was added for mercury stabilization. Quantification used an Agilent 8900 ICP-MS with an integrated autosampler and MassHunter 5.2 software, operating in helium collision-gas kinetic-energy-discrimination mode. Quality control used NIST SRMs 1548a, 1568b, 1643f and, for speciation work, NMIJ CRM 7503a/7405b and NIST SRM 2976.
Arsenic speciation was targeted rather than universal. Rice-containing dry foods/supplements with total As above 50 µg/kg were analyzed by FDA EAM 4.11 using hot-block extraction and HPLC-ICP-MS for inorganic arsenic. The two fish-containing puree/mixture products were analyzed for arsenic species by a seafood/seaweed method using hot-block extraction with anion/cation exchange HPLC-ICP-MS. The same two fish-containing products were analyzed for methylmercury using double liquid-liquid extraction followed by direct mercury analysis; the Met-Hg LOD was 4 µg/kg. Predictor screening used JMP Pro v17 with Bootstrap Forest / model screening, 10-fold cross-validation, post-hoc Pearson correlations, Bonferroni-adjusted alpha of 0.001, and effect-size estimates from correlation coefficients.
Implications
Certification (HMTc): This is high-value A-tier occurrence evidence for U.S. infant and young-child foods because it uses FDA methods, low detection limits, and a large 2023 sample frame that complements FDA’s 2021 puree survey and FY2023 lead-guidance analytical table. It contributes category-level occurrence signals for rice-containing dry foods/snacks, puree/mixtures, root-vegetable purees, fish-containing baby foods, and baby/young-child liquids without supporting brand-level comparison or any standalone certification-limit value.
Courses: Strong teaching case for how FDA pairs occurrence surveillance with ingredient-predictor screening, including the limits of weak-to-modest R2 models and the need to distinguish total As from inorganic As, total Hg from methylmercury, and ingredient association from causation.
App: Adds U.S. 2023 ingredient-linked occurrence evidence for rice/tAs, sweet potato-carrot/root-vegetable/Pb, tomato-spinach-carrot/Cd, fish/tHg, and kale/broccoli/brassica/Tl signals. Downstream app use should retain the study’s matrix grouping and its warning that geographic origin, soil conditions, and processing also influence final levels.
Microbiome: No direct microbiome endpoint.
Wiki pages this source may touch
- arsenic-total
- arsenic-inorganic
- cadmium
- mercury-total
- mercury-methyl
- lead
- thallium
- rice
- apple-juice
- pear
- root-vegetables
- sweet-potato
- carrots
- tomato
- spinach
- leafy-vegetables
- broccoli
- fish
- infant-and-child-foods-master
- baby-cereals-dry-rice-based
- baby-cereals-dry-non-rice
- teething-and-snacks-rice-based
- teething-and-snacks-non-rice
- fruit-purees
- root-vegetable-purees
- non-root-vegetable-purees
- mixed-meals-rice-containing
- mixed-meals-non-rice
- fish-containing-baby-foods
- infant-formula-powder
- fruit-juices-apple-containing
- fruit-juices-non-apple
- fda-closer-to-zero
- fda-iAs-rice-cereal-100ppb
- fda2025-lead-processed-baby-foods
Verification notes
- Ad hoc RGSA manual-fetch ingest by Codex on 2026-05-18 from
raw/Research Gate sent by author/Barberetal_FoodAdditContam2025.pdf. - Brand firewall: Table 1 reports product counts by named manufacturers/private labels. Those names are intentionally omitted here because they are sampled-product brand identifiers, not method vendors or regulatory-event subjects.
- Thallium arithmetic caveat: the Results text reports “12 of 394 (24%)” puree/mixture products above LOQ. Since 12/394 is not 24%, this page preserves the numerator/denominator and flags the parenthetical percentage as a source-side arithmetic inconsistency.
- Matrix vocabulary note:
fruit-juiceis used here for the source’s liquids/drinks subset because the standard matrix list does not yet contain a better baby/young-child juice term. - Fresh-context audit (Codex/Meitner, 2026-05-18) returned REVISE only for the unflagged
fruit-juicematrix term; this verification note resolves the vocabulary concern. Numerical fidelity, methods/speciation, brand firewall, and wiki/HMTc firewall were clean.
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.