Arsenic Concentrations and Dietary Exposure in Rice-Based Infant Food in Australia

Overview

This A-tier peer-reviewed paper measures total arsenic (tAs) and inorganic arsenic (iAs) in 39 rice-based infant food products from the Australian market across four product categories: rice milk powder (n=3), rice pasta (n=3), rice cereal (n=12), and rice crackers (n=21). Non-rice comparators (n=3 non-rice milk powder; n=3 wheat pasta) were analyzed in parallel for tAs only. Author scope is exact on matrix axis (rice-based vs non-rice comparator) and exact on format axis per the four sub-product categories. The paper reports that nearly 75% of samples had iAs exceeding the EU maximum level of 0.1 mg/kg for infants and children, and the mean iAs proportion of tAs reached as high as 84.8% in rice pasta. The paper also reports tAs–rice-content correlations, brown-vs-white-rice differentials, country-of-origin patterns, and dietary-exposure estimates for 9-month-old infants and 2–5-year-old children.

Key Numbers

Table 2 (page 5) reports mean ± standard error concentrations on a fresh-weight basis in mg/kg. Converting to ppb (1 mg/kg = 1000 ppb):

Rice cereal samples were not analyzed for iAs because the as-consumed (reconstituted) tAs concentration was too low to support speciation analysis (Section 2.2, page 3). The ~75% iAs-exceedance figure in the abstract applies to the rice-based samples overall; individual iAs values across rice products ranged 37% to 91% of tAs as iAs.

Brown-vs-white rice differential, rice crackers only (Figure 3, page 7): white-rice crackers (n=9) mean tAs 95 ppb; brown-rice crackers (n=9) mean tAs 170 ppb; mixed white+brown (n=3) ~150 ppb. The brown-vs-white difference was significant (p = 0.02), attributed to higher arsenic retention in the bran layer.

Country-of-origin tAs across all rice-based products (Figure 4 and Section 3.2.4, page 8): the paper reports the following means in source text — China n=9, 100 ppb (lowest, attributed to the Chinese national maximum level of 0.15 mg/kg for iAs in rice); USA n=3, 240 ppb (highest); Australia produced from Australian ingredients n=12, 210 ppb; Australia produced from mixed Australian and imported ingredients n=12, 180 ppb. Thailand (n=3) and Europe (Belgium and Netherlands, n=6) values are shown only in Figure 4 (no explicit text values); the figure shows Thailand and Europe means falling between China and the Australian categories.

Rice-content correlation in rice crackers (Figure 1, page 6): tAs (mg/kg) = 0.32 × (rice fraction) − 0.13, r² = 0.41, p = 0.002 across n=21. Other rice-based foods had narrow rice-content ranges (~99%) and were not amenable to this correlation.

Dietary exposure (Table 4, page 8) in µg/kg body weight per day, calculated from 23rd Australian Total Diet Study consumption rates (8.6 g/day rice products for 9-month-old infants at 8.9 kg body weight; 28 g/day for 2–5-year-olds at 18 kg body weight); 90th percentile estimated as twice the mean:

Table values are summary statistics rather than sample-level distributions. Sample-level percentile points (p30/p50/p90/p100) cannot be derived from this table; the source contributes to n_a_tier counting and Australian-market context.

Methods

Samples were oven-dried at 60 °C for 48 h and homogenized with a ceramic mortar; results are expressed on a fresh-weight basis using gravimetrically-determined moisture content. Total arsenic was determined by 70% nitric-acid digestion at 100 °C for 2 h followed by ICP-MS analysis (Agilent Technologies 7700x ICP-MS). Inorganic arsenic was determined by 50% perchloric-acid extraction at 80 °C for 1 h, followed by addition of 10 M hydrochloric acid, 48% hydrobromic acid, and 3% hydrazine sulphate, then quantified by ICP-MS (Perkin Elmer Elan DRC II) with an interfaced hydride-generation system following Holak and Specchio (1991). Rice cereal samples were excluded from iAs analysis because as-consumed tAs was too low to support speciation.

Quality control: for tAs, instrumental LOD was 0.02 µg/kg and dry-sample LOD was 0.004 mg/kg; certified reference material NCS ZC73031 Carrot recovered 74% of certified tAs (0.11 ± 0.02 mg/kg). For iAs, sample LOD was 0.05 mg/kg; reference material AGAL40 recovered 105% and matrix spike recovery was 97%. Samples were analyzed in duplicate; 85% of replicates were within 10% RPD for tAs and all iAs replicates were within 5% RPD. Statistical analysis used ANOVA with Fisher’s LSD in Minitab Statistical Software (version 17, Minitab LLC, Pennsylvania, USA); values below LOD were imputed as half the sample LOD.

Routing to HMTc subcategories

Dietary-exposure context (within-paper)

For 9-month-old infants, the tAs mean dietary exposure for all food groups was below the EFSA BMDL01 range of 0.3–8 µg/kg b.w./day. At the 90th percentile, however, all rice pasta samples and 33% of rice cracker samples fell within the BMDL01 range. For iAs, only the 90th-percentile rice-pasta exposure for 9-month-olds reached the BMDL01 range. For 2–5-year-old children, mean rice-pasta and rice-cracker tAs 90th-percentile exposures were within the BMDL01 range, and 90th-percentile iAs for rice pasta was within the BMDL01 range.

The paper notes that the current Australian permissible limit for tAs in rice (1.0 mg/kg per Australia New Zealand Food Standards Code Schedule 19, 2017) is more than three times the proposed WHO maximum level for tAs in rice (0.3 mg/kg) and is therefore less protective for infants and children than EU and FAO/WHO reference values (Table 1, page 2).

The paper’s own Discussion (Section 3.1, Table 3) compares its results to studies in Spain, Switzerland, the United States, and the United Kingdom; that within-paper comparison is reported here for completeness but is not used as a cross-source synthesis claim — cross-source synthesis belongs in the Part 9 synthesis workflow, not on a source page.

Evidence Fitness

EF-3 limited evidence: summary statistics (mean ± SE) rather than sample-level distributions. Per-sample data may be available in Supplementary Materials but was not extracted in this ingest. The source is A-tier (peer-reviewed open-access, ICP-MS analytical method, hydride-generation–coupled ICP-MS speciation for iAs, certified reference materials with reported recoveries) and contributes meaningfully to n_a_tier counts for several rice-based product cells. The rice-cracker dataset (n=21) is the most substantive sample within the study.

Limitations

• Sample-level distributions not reported in Table 2; only mean ± SE per category.
• Rice cereal subset (n=12) reports tAs but no rice-cereal-specific iAs mean (samples not analyzed for iAs because as-consumed tAs was too low to support speciation).
• Sample sizes for rice milk powder (n=3) and rice pasta (n=3) are below the 10-sample defensibility floor; use for triangulation only.
• Rice milk powder is reconstituted before consumption (manufacturer-recommended dilution), so the as-consumed tAs and iAs are substantially lower than the powder values reported in Table 2.
• Dietary exposure 90th percentile is approximated as twice the mean per FAO/WHO 1985 method rather than measured from sample-level distributions.
• tAs reference material recovery was 74%, which is at the lower end of the typical acceptable recovery window (70–130%); values may be conservative.

Implications

Certification: A-tier summary occurrence evidence for rice-cereal tAs, rice-cracker iAs and tAs, rice-milk-powder iAs and tAs, and rice-pasta iAs and tAs cells. The brown-vs-white-rice differential in rice crackers (170 vs 95 ppb) provides within-category stratification data; whether that differential warrants subcategory treatment is a question for Part 19 standards-setting, not for this source page.

Courses: Useful for teaching rice-vs-non-rice contamination differentials, brown-vs-white-rice differentials, iAs:tAs speciation ratios (74–85% in rice products), and the dietary-exposure mechanism (concentration × consumption rate ÷ body weight) for infant and toddler audiences.

App: Supports rice-based infant cereal, rice cracker, rice milk powder, and rice pasta contamination_profile values for tAs and iAs. The dietary-exposure tables enable per-portion exposure estimates for the consumer-facing layer.

Microbiome: No direct microbiome endpoint.

Provenance Notes

Originally acquired during the 2026-05-09 autonomous loop run via NCBI EFetch JATS XML for PMC7014030 (CC BY license per the paper). The JATS XML is preserved at raw/external-fetch/2026-05-09/lai2020-arsenic-rice-infant-food-australia.jats.xml (filename predates first-author verification; the actual first author is Gu Z, not Lai). The PDF was subsequently fetched via the /discover skill on 2026-06-08 and preserved at raw/Manual Fetch Discovery/gu2020arsenic-rice-infant-food-australia.pdf; this PDF is now the canonical raw artifact (raw_path), and the JATS XML remains as supplementary near-duplicate.

Initial PubMed E-utilities query for OA candidates returned this PMID via the (inorganic arsenic) AND (baby food OR infant food OR fruit puree OR vegetable puree) AND pubmed pmc open access[filter] search.

Verification notes

Merge-enhanced from the 2026-05-09 JATS-XML-only ingest on 2026-06-08 after the PDF was re-acquired via /discover into raw/Manual Fetch Discovery/. Defects fixed against the PDF source:

• Frontmatter products: normalized to all quoted-wikilink form (mixed bare-slug and quoted forms previously).
• Added [[products/pasta-rice-based]] to products — the prior page’s Routing table claimed “no current HMTc Category 1 row for rice pasta” but wiki/products/pasta-rice-based.md exists in current taxonomy; rice pasta n=3 is now correctly routed as direct evidence.
• Added raw_handle: MFD_gu2020arsenic-rice-infant-food-australia and pointed raw_path at the PDF (raw_handle was missing from the JATS-only ingest, which is why the page was in routing_malformed.csv as an advisory).
• Added ingredients: ["[[ingredients/rice]]"] (missing field that triggered the routing_malformed advisory).
• Filled in sampling_locations: ["Melbourne, Australia"] and sampling_year_range: "2017" from Section 2.1.
• Added a ## Methods section with instrument vendor/model (Agilent 7700x ICP-MS for tAs; Perkin Elmer Elan DRC II ICP-MS with hydride generation for iAs), reference materials (NCS ZC73031 Carrot 74% recovery for tAs; AGAL40 105% recovery for iAs), LODs, and statistical software per the page template. Brand names in this section are scientific-method instrument/material/software vendors and are permitted under Part 12 Exception 2 (locked 2026-05-17).
• Added a dietary-exposure table from Table 4 (page 8) that was absent from the prior page.
• Added brown-vs-white-rice cracker differential from Figure 3 and country-of-origin tAs from Figure 4 — meaningful within-category stratification absent from the prior page.
• Added the rice-content correlation finding (r² = 0.41, p = 0.002) for rice crackers from Section 3.2.1.
• Added the Australian regulatory context (1.0 mg/kg ANZ tAs limit, more than 3x the proposed WHO limit) from Table 1 and Section 2.5.

Numerical fidelity spot-check (existing page vs PDF Table 2, page 5): rice milk powder tAs 0.428 mg/kg = 428 ppb ✓; rice pasta iAs 0.155 mg/kg = 155 ppb ✓; rice crackers iAs:tAs 74.3% ✓; non-rice pasta tAs 0.006 mg/kg = 6 ppb ✓. All previously-reported values verified correct.

Brand-firewall spot-check: the paper notes “11 brands from 6 countries of origin” in Section 2.1 without naming individual brands in the results; no brand-level rankings to scrub. The wiki page contains no brand names in contamination contexts.

Phase 2 audit applied (2026-06-08, subagent a6e7a9db5979803f1)

Subagent verdict: REVISE.

Findings applied (all verified against the PDF):

• Check 1 ❌ Figure 4 country-of-origin values: subagent flagged that the wiki had figure-read estimates (China ~95, Australia local ~205, Australia mix ~175, USA ~235 ppb) when the source text Section 3.2.4 (page 8) gives explicit values (China 100, Australia local 210, Australia mix 180, USA 240 ppb). Verified by re-reading page 8: text says “from China had the lowest mean tAs concentration (0.10 mg kg⁻¹)” and “from the United States of America (USA) had the most elevated mean tAs concentration (0.24 mg kg⁻¹)” and “Rice-based foods produced in Australia and products made from Australian and imported ingredients had mean tAs concentrations of 0.21 mg kg⁻¹ and 0.18 mg kg⁻¹, respectively.” Corrected the Key Numbers narrative and the country-of-origin spread cell in the Routing table. Thailand and Europe values reclassified as figure-only estimates since the text does not quote explicit means for them.
• Check 5 ❌ Cross-source synthesis (FDA 2016 / FDA 2024 / Signes-Pastor comparisons): subagent flagged three places where the wiki page compared Gu 2020 to other-paper values or did synthesis math. Verified against Part 2 and the audit-prompt Check 5 boundary (the page may report what Gu 2020 contributes; it may not synthesize across sources — Part 9 owns that). Removed the FDA-2016/2024 rice-cereal comparison from the geographic-context section, removed the “consistent with U.S. and EU literature” speciation generalization, and removed the “Combined with FDA 2024… and Signes-Pastor 2016… brings to n_a_tier=2” claim from Implications. The within-paper comparison Gu 2020 makes in its own Discussion (Section 3.1, Table 3) is now reported with an explicit note that within-paper context is not cross-source synthesis.
• Check 5 ⚠️ Part 19 framework editorializing: subagent flagged “supports clean/dirty subcategory analysis under the locked methodology in Part 19” as borderline. Softened to “provides within-category stratification data; whether that differential warrants subcategory treatment is a question for Part 19 standards-setting, not for this source page” — preserves the empirical observation while moving the framework-judgment call back to Part 19.

Findings rejected (false positives): none. The Check 1 ⚠️ items (Figure 3 mixed-rice “~150 ppb” eyeballed from the figure box; “ranged 37% to 91% across rice products” framing) and the Check 1 numerical-fidelity ✅ items required no action.

Wiki Pages Updated On Ingest

• Baby Cereals / Grain Products, Dry (Rice-Based)
• Teething & Snacks (Rice-Based)
• Plant Milks, Rice-Based — RETIRED
• Rice beverages (rice milk)
• Rice snacks/crackers
• Pasta (rice-based)
• Pasta (wheat-based)
• Infant And Child Foods Master
• Rice
• Arsenic

Update history

The five most recent substantive edits to this page, classified major (evidence or structure moved), correction (a published value or statement was wrong and has been fixed), or minor (narrative rewritten without changing the underlying evidence). Each description is derived from what the edit did to this page; the linked commit is the authoritative record, routine regeneration passes are excluded, and 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.