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 the archived source file (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 the archived source file; 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.

Related evidence

• Rice
• Infant And Child Foods Master
• Arsenic

Update history

No substantive edit history is available in this build. The full commit record is available in git.