Noh et al. 2023 — Arsenic species in rice-based processed products in South Korean markets

Cross-sectional market-survey monitoring study of total arsenic (tAs) and four arsenic species (AsV, AsIII, MMA, DMA) in 239 rice-based processed foods purchased from South Korean domestic markets between February and August 2019. The authors stratified samples into ten product categories (seven adult, three for infants and young children) and ran a margin-of-exposure (MOE) risk assessment against the EFSA BMDL₀₁ of 0.3 µg iAs/kg bw/day across five Korean age strata. Inorganic arsenic (iAs) was computed as the sum of AsV and AsIII. The paper concludes that mean iAs is broadly comparable to international rice-based-food monitoring data; that rice cakes, adult snacks, and rice-based baby powder represent the categories with MOE values low enough to flag for age-specific concern; and that the iAs/tAs correlation is modulated by ingredient diversity (seaweed-containing categories show low iAs/tAs because organoarsenicals from seaweed inflate tAs without inflating iAs).

Key numbers

All values are µg/kg wet-weight on the as-marketed product (HMR rice, rice cake, etc. are sold cooked or shelf-stable; powders sold as dry product). Tables 2, 3, 4, and 5 anchor the figures below; page numbers reference Food Science and Biotechnology 32:1361–1372.

• Total arsenic (tAs, mean ± SD µg/kg; detection frequency %), Table 2 page 1366:
  • HMR rice (n=30): 110 ± 150 (97%)
  • Rice cake (n=30): 69 ± 33 (97%)
  • Porridge adult (n=25): 49 ± 43 (72%)
  • Noodle (n=33): 180 ± 190 (94%)
  • Bread (n=20): 40 ± 30 (90%)
  • Snack adult (n=45): 120 ± 150 (84%)
  • Powder adult (n=19): 160 ± 170 (100%)
  • Snack for baby (n=14): 110 ± 80 (100%)
  • Porridge for baby (n=14): 22 ± 40 (29%)
  • Powder for baby (n=9): 160 ± 70 (100%)
• Inorganic arsenic (iAs = AsV + AsIII; mean ± SD µg/kg; detection frequency %), Table 2 page 1366:
  • HMR rice: 31 ± 18 (97%)
  • Rice cake: 37 ± 18 (97%)
  • Porridge adult: 19 ± 22 (72%)
  • Noodle: 48 ± 34 (93%)
  • Bread: 23 ± 16 (90%)
  • Snack adult: 43 ± 36 (82%)
  • Powder adult: 85 ± 56 (100%)
  • Snack for baby: 62 ± 55 (100%)
  • Porridge for baby: 4.4 ± 8 (29%)
  • Powder for baby: 74 ± 31 (100%)
• iAs range and iAs/tAs ratio by category, Table 3 page 1367 (this study columns only):
  • HMR rice: ND–90.5 µg/kg, mean 30.9, iAs/tAs 29%
  • Rice cake: ND–87.2, mean 36.6, 53%
  • Porridge: ND–73.0, mean 18.8, 39%
  • Noodle: ND–150, mean 48, 26%
  • Bread: ND–46.4, mean 22.7, 57%
  • Snack: ND–144, mean 43.2, 36%
  • Powder: ND–283, mean 84.7, 52%
• tAs–iAs correlation factor r and ingredient diversity index H′ by category, Table 4 page 1369 (asterisks mark correlations significant at the t-test p<0.05 threshold the authors used):
  • HMR rice: r=0.0657, p=7.30×10⁻¹, H′=3.45 (no significant correlation)
  • Adult snack: r=0.280, p=6.23×10⁻², H′=3.23 (no significant correlation)
  • Noodle: r=0.311, p=7.81×10⁻², H′=2.87 (no significant correlation)
  • Porridge adult: r=0.565, p=3.23×10⁻³, H′=2.95
  • Rice cake: r=0.781, p=3.56×10⁻⁷, H′=2.88
  • Bread: r=0.823, p=8.47×10⁻⁶, H′=2.29
  • Porridge for baby: r=0.829, p=2.51×10⁻⁴, H′=2.85
  • Powder for baby: r=0.896, p=1.08×10⁻³, H′=2.03
  • Powder adult: r=0.925, p=1.54×10⁻⁸, H′=1.97
  • Snack for baby: r=0.990, p=1.20×10⁻¹¹, H′=1.84
• Risk assessment (MOE = BMDL₀₁ / EDI; BMDL₀₁ = 0.3 µg iAs/kg bw/day; MOE < 100 = concern), Table 5 page 1369. Reference body weights: age 1–2 = 12.0 kg, age 3–5 = 17.6 kg, age 6–18 = 45.5 kg, age 19–64 = 64.5 kg, age ≥65 = 60.6 kg (KNHANES 7th edition 2016–2018).
  • Age 6–18: rice cake MOE 24, adult snack MOE 37.1, bread MOE 193, noodle MOE 564, adult powder MOE 531, HMR rice MOE 8 460, all categories combined MOE 1 920.
  • Age 19–64: rice cake MOE 30.3, adult snack MOE 162, adult powder MOE 226, noodle MOE 457, bread MOE 4 760, HMR rice MOE 16 000, all combined MOE 27 100.
  • Age ≥65: rice cake MOE 31.2, adult snack MOE 259, adult powder MOE 252, noodle MOE 1 050, bread MOE 6 840, HMR rice MOE 94 000, all combined MOE 137 000.
  • Age 1–2: rice-based powder for baby MOE 29.3, porridge for baby MOE 35 700, snack for baby MOE 64 900, all combined MOE 33 600.
  • Age 3–5: rice-based powder for baby MOE 1 430, porridge for baby MOE 5 510, all combined MOE 3 470.
• Analytical performance, Table 1 page 1365: tAs R²=0.9961, MDL 0.32 µg/kg, MQL 1.03 µg/kg; per-species MDL 0.26–0.37 µg/kg; interday accuracy 94.9–103% and interday precision 0.94–8.2% RSD across tAs, iAs, MMA, DMA. CRM NIST SRM 1568b rice flour (certified tAs 285 µg/kg, iAs 92 µg/kg, MMA 11.6 µg/kg, DMA 180 µg/kg) used for accuracy validation.
• Diet-driven speciation findings, page 1367–1368: husked rice noodles had significantly (p<0.05) higher iAs than polished rice noodles (Figure 1F); seaweed-containing noodles had significantly (p<0.05) higher tAs than seaweed-free noodles (Figure 1C) without a comparable iAs shift, supporting the authors’ interpretation that the seaweed organoarsenicals inflate tAs but not iAs.

Methods (brief)

Samples were purchased from offline retailers and online markets in South Korea between February and August 2019. The ten categories were chosen by sales volume and purchasing preference per Kim et al. 2018 and Han & Gouk 2014, with per-category sample counts set proportional to category sales volume. Total arsenic was digested per Korean Food Code (KFC) method 8.9.1.4: 0.5 g of dried, powdered sample microwave-digested in 4 mL of 70% HNO₃ plus 1 mL of 30% H₂O₂ in PTFE vessels (ETHOS EASY, Milestone, Italy; five-stage temperature ramp to 190 °C). Speciation extraction followed KFC method 8.9.1.5: 1.0 g of sample heated in 5 mL of 1% HNO₃ at 90 °C for 90 min, centrifuged at 3000×g, filtered through 0.45 µm nylon. tAs quantified by ICP-MS (Agilent 7700x with He collision cell at 4 mL/min, m/z 75). Speciation by HPLC-ICP-MS using a CAPCELL PAK C18 MG column (4.6×250 mm, 5 µm; Shiseido) with mobile phase 0.05% methanol / 10 mM sodium 1-butane sulfonate / 4 mM malonic acid / 4 mM tetramethylammonium hydroxide (pH 2.7, nitric acid) at 0.75 mL/min, 25 °C, 15 µL injection. iAs was computed as AsV + AsIII; AsV and AsIII were not independently certified in NIST SRM 1568b, so the manufacturer’s combined AsV+AsIII value (92 µg/kg) was used for the iAs accuracy check. Non-detect handling differs between the two reporting layers in the paper: Table 2 footnote d sets Results < MDL to 0 for the category-level mean ± SD figures; the EDI / MOE risk assessment on page 1364 sets ND equal to the iAs MDL of 0.73 µg/kg before multiplying by category-specific intake, which is the more conservative convention for the carcinogenic-risk MOE calculation. Daily iAs intake was computed as (daily category-specific consumption × mean iAs concentration with ND set to MDL) / mean body weight, using KNHANES 7th edition (2016–2018) intake data and the conservative lower-bound BMDL₀₁ of 0.3 µg iAs/kg bw/day from EFSA 2009 (Alexander et al. 2009). Statistical comparisons used t-test and one-way ANOVA at α=0.05 in R; an ingredient diversity index H′ = −Σ pᵢ ln pᵢ was computed per category to test whether non-rice ingredients drive the tAs–iAs decoupling. Grant funding: Korean Ministry of Food and Drug Safety (15162MFDS077) and Korea University internal grant K2207571; authors declare no conflicts of interest.

Implications

The paper contributes A-tier occurrence data for the rice-bulk-grain, rice-cake (rice-snacks-crackers), rice-noodle, rice-bread (bread-and-baked-goods), adult-snack, baby-rice-cereal (baby-cereals-dry-rice-based), baby-rice-snack (teething-and-snacks-rice-based), and rice-based baby porridge (infant-porridges-dairy-free) routes for both iAs and tAs in a Korean (East Asian) jurisdiction. It is the second-largest single-study Korean dataset on speciated arsenic in rice-derived processed foods after Lee et al. 2018 (polished-rice survey, cited inline), and the first published Korean monitoring effort to extend speciated arsenic measurement across noodles, bread, and the full HMR / baby-food product slate using a single validated method. The three operationally important findings for downstream HMTc work are (1) the rice-based powder for baby category has both high mean iAs (74 µg/kg) and the lowest MOE among baby products (29.3 at age 1–2), placing it in the same age-1–2 MOE-below-100 band as the rice cake category for older children and adults; (2) the iAs/tAs ratio varies roughly twofold across categories (26% in noodle to 57% in bread) and correlates inversely with the per-category ingredient diversity index H′ — seaweed-containing categories inflate tAs without inflating iAs, so any tAs-only measurement in these categories overstates the inorganic burden; (3) within the noodle and snack categories, husked rice samples have significantly higher iAs than polished rice samples (Figure 1B,F), and seaweed-containing noodle samples have significantly higher tAs than seaweed-free noodle samples (Figure 1C). The complete per-category mean ± SD with per-species detection frequencies provides occurrence data for the rice-bulk-grain, rice-snacks-crackers, rice-noodles, bread-and-baked-goods, baby-cereals-dry-rice-based, teething-and-snacks-rice-based, and infant-porridges-dairy-free routes.

Verification notes

• 2026-06-09 fresh ingest from raw/Manual Fetch Discovery/noh2023-arsenic-rice-snacks-korea.pdf (Karen-fetched via the /discover skill into the Manual Fetch Discovery folder). Three identity checks all negative: no existing wiki source page matches DOI 10.1007/s10068-023-01270-9, raw_handle MFD_noh2023-arsenic-rice-snacks-korea, or cite-key noh2023. NEW path.
• Sample-size cross-check: the abstract gives a seven-category breakdown (rice n=30; rice cake n=30; porridge n=39; noodles n=33; bread n=20; snack n=59; powder n=28; total 239); Materials and methods plus Table 2 give a ten-category breakdown (HMR rice n=30; rice cake n=30; porridge adult n=25; noodle n=33; bread n=20; snack adult n=45; powder adult n=19; snack for baby n=14; porridge for baby n=14; powder for baby n=9; total 239). The two presentations are arithmetically consistent — the abstract pre-aggregated baby and adult porridge (25+14=39), snack (45+14=59), and powder (19+9=28). Wiki page uses the ten-category breakdown because it is the basis for all downstream tables and the MOE risk assessment.
• Speciation discipline (Part 14 / audit Check 3): the paper measures both tAs (via microwave digestion + ICP-MS) and iAs (via HPLC-ICP-MS speciation, iAs = AsV + AsIII). Both tAs and iAs are correct in metals:. AsV and AsIII are reported separately in Table 2 but are not separate metal slugs in the wiki taxonomy; they are subsumed under iAs. MMA and DMA are reported but the wiki does not currently track organoarsenicals as a separate metals slug, so they are documented in Key numbers and Methods but not in metals:.
• “Powder adult” product slug: the paper’s adult powder category is a Korean retail category of rice-based powder mixes (often used in beverages, breakfast drinks, or instant porridge preparation). No existing wiki product slug is a clean fit. Routed via rice-bulk-grain as the closest broad-scope rice-product slug; this matches the system-prompt rule “underspecifying is safer than overspecifying” — downstream synthesis or a future Step 0 Lock review may want a dedicated rice-flour-powders or similar slug, but that is not this ingest’s call.
• “Porridge for baby” routing: routed to infant-porridges-dairy-free rather than infant-porridges-dairy because rice-based Korean baby porridge is the rice-grain-and-water (mieumi) form, not a dairy-containing porridge. The paper does not specify whether any of the 14 baby porridge samples contain dairy; the slug call is the safer dairy-free assumption pending paper-level disclosure.
• Brand firewall (Part 12, strict 2026-05-17): the paper reports category-level mean ± SD across n samples without naming any brand of HMR rice, rice cake, noodle, etc. No brand-firewall content to strip. Methods-section vendor names (Agilent 7700x, Shiseido CAPCELL PAK C18 MG column, ETHOS EASY Milestone microwave digester, NIST SRM 1568b rice flour, R statistical software) are scientific-method vendor names and are preserved per the 2026-05-17 Exception 2 carve-out.
• License: the published article is © Korean Society of Food Science and Technology 2023, published by Springer Nature under a standard publisher-copyright (subscription) agreement. No open-access CC license declared. license: publisher-copyright reflects that posture.
• 2026-06-09 audit-application (fresh-context subagent verdict REVISE): two findings applied. (1) Check 1 ⚠️ — Methods section had conflated non-detect handling. Verified against page 1364: the EDI / MOE risk assessment uses ND set to the iAs MDL of 0.73 µg/kg (per the Eq. 3 narrative), whereas Table 2 footnote d uses ND set to 0 only for the category-level mean ± SD reporting. Methods paragraph rewritten to distinguish the two reporting layers. (2) Check 5 ⚠️ — Implications section had two sentences that drifted from “what the paper found” toward HMTc methodology prescription (instructing how threshold-setting should treat the H′ covariate, and asserting workbench pool-admission readiness). Both rewritten as observations: the iAs/tAs ratio finding now reports the observed inverse correlation and the operational caveat that tAs-only measurement overstates inorganic burden in seaweed-containing categories, without prescribing a conversion factor; the pool-admission sentence is rewritten as “provides occurrence data for [the seven routes]” without preempting workbench admission. Verdict moves to REVISE-applied; downstream synthesis remains the Part 9 workflow’s job.

Wiki pages this source may touch

• arsenic-inorganic
• arsenic-total
• rice
• seaweed
• rice-bulk-grain
• rice-snacks-crackers
• rice-noodles
• bread-and-baked-goods
• mixed-meals-rice-containing
• baby-cereals-dry-rice-based
• teething-and-snacks-rice-based
• infant-porridges-dairy-free

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.