Han et al. 2024 — Nickel occurrence and dietary exposure in Zhejiang Province, China

This study quantifies nickel concentrations across six food categories in Zhejiang Province, China (2628 samples, 2018–2019) and assesses dietary exposure risk by age group. Beans (n=5) carry exceptionally high nickel levels (mean 3.094 mg/kg, all five samples above the GB 2762-2022 1 mg/kg national limit), while cereal products and fruits stay below the limit. Children aged 0–6 are the only population segment showing unacceptable cumulative exposure (THQ = 1.078) under high-consumption/high-contamination assumptions.

Key numbers

Sampling: 11 cities in Zhejiang Province, 2018–2019, multistage random sampling at the breeding/acquisition (field) and intermediate (market, store, online commercial) stages. Sample counts by category — Section 2.1 (Sampling) totals vs Section 3.1 (Sample Characteristics) / Table 1 totals: beans n=5 (both); aquatic food and products n=1404 (§2.1) vs n=1405 (§3.1 + Table 1); meat and products n=691 (§2.1) vs n=690 (§3.1 + Table 1); vegetables and vegetable products n=392 (§2.1) vs n=365 (§3.1 + Table 1); cereal products n=27 (both); fruits n=145 (§2.1) vs n=136 (§3.1 + Table 1). The Key numbers table above uses §3.1 / Table 1 values throughout (the granular per-subcategory source). Body weights by age group from the consumption survey:

Population census shares reported for the consumption survey (three-bucket grouping): 16.60% aged 0–14, 70.14% aged 15–59, and 13.26% aged ≥60.

Daily Ni intake (µg/kg-bw/d), Mode D (P95 concentration × P95 consumption), by age group:

Beans and vegetables/vegetable-products were the main sources of dietary Ni exposure (>50% of total for low-consumption modes A and C); vegetables dominate in high-consumption modes B and D. USEPA reference dose used: 20 µg/kg-bw/d. The paper also references the EFSA Tolerable Daily Intake — initially 2.8 µg/kg-bw in 2015 and later modified to 13 µg/kg-bw — as a comparator; the 0–6 Mode D total intake of 21.57 µg/kg-bw/d exceeds both EFSA TDI revisions and the USEPA RfD.

Methods (brief)

Inductively Coupled Plasma Mass Spectrometry on a NexION 300D ICP-MS (PerkinElmer, Waltham, MA, USA) equipped with a concentric nebulizer and collision reaction cell. Samples (0.5 g) were microwave-digested in nitric acid (6–8 mL; chromatographic purity, Merck, Darmstadt, Germany) with hydrogen peroxide (1 mL) added for hard-to-digest matrices in a Mars-6 microwave digestion system (CEM, Charlotte, NC, USA), following a three-step program (Step 1: 120 °C, 5 min ramp + 5 min isothermal; Step 2: 150 °C, 5 min ramp + 10 min isothermal; Step 3: 190 °C, 5 min ramp + 25 min isothermal). Digested solution was diluted to 25 mL with ultrapure water. A Ni standard stock and tuning solution (Li, Y, Ce, Tl, Co; 10 ng/mL) were sourced from CNW Technologies GmbH (Düsseldorf, Germany). LOD: 3.5 µg/kg for vegetables and fruits, 3.0 µg/kg for other matrices. Samples below LOD set to ½ LOD. Quality control: one QC sample per ten samples; certified reference materials; standard-addition recovery. Total nickel only — no speciation between Ni(II), Ni(0), or Ni carbonyls (metals tag: Ni).

Risk assessment used the USEPA target hazard quotient method (EDI = C × D × T / BW; HQ = EDI / RfD; RfD = 20 µg/kg-bw/d). Cumulative risk across food categories was expressed as THQ = ΣHQ. Four consumption modes A–D corresponded to median/P95 combinations of concentration × consumption (A: P50×P50; B: P50×P95; C: P95×P50; D: P95×P95). Statistical analyses: chi-squared test for detection/overstandard rates; Anderson–Darling normality test; Wilcoxon rank-sum test for skewed distributions. Analyses performed in Excel and R 4.2.2.

Implications

Certification: Beans are a high-Ni outlier among the food categories sampled in this study — all five sampled bean samples (one soybean, two mung bean, two ormosia) exceeded the GB 2762-2022 1 mg/kg national standard, though the very small n (5) limits the strength of inference for the bean category as a whole. Within the larger aquatic food sample (n=1405), canned fish (n=82; 13.41% overstandard rate, P95 1.539 mg/kg) and shellfish (n=50; 12% overstandard, P95 1.279 mg/kg) were the highest-Ni sub-categories. This dataset contributes occurrence and exposure data for legumes, canned fish, shellfish, and processed meats relevant to Cat 1 (infant/child foods) Ni exposure modelling, including the finding that the 0–6 year THQ under high-concentration / high-consumption assumptions (Mode D) was 1.078, the only age group exceeding the HQ=1 threshold in any consumption mode.

Courses: This paper illustrates how ICP-MS occurrence data combine with consumption surveys to produce age-stratified dietary exposure estimates, and how the THQ method aggregates risk across multiple food categories. It is also a useful teaching example of internal-discrepancy hygiene — Section 3.1 text totals (n=1404 aquatic, n=691 meat, n=392 vegetables, n=145 fruits) differ from Table 1 totals (1405, 690, 365, 136); the wiki preserves both with the discrepancy noted.

App: Beans, canned fish, and shellfish contribute disproportionately to per-serving Ni in this Zhejiang Province sample. Per-ingredient Ni occurrence inputs for East-Asian dietary contexts can draw on the P50/P95 values reported here, subject to the small bean n.

Microbiome: Not applicable — paper does not address gut microbiome effects of Ni exposure.

Wiki pages this source may touch

• nickel
• beans
• legumes
• vegetables
• non-root-vegetables
• leafy-vegetables
• cereals
• fruits
• meat
• meat-and-poultry
• seafood
• freshwater-fish
• shellfish
• molluscs
• canned-fish
• processed-meats
• china-gb-2762-2022-contaminants

Verification notes

• 2026-05-29: Manual-fetch re-ingest pass against the source PDF (raw/Manual Resourcing/Occurrence and Exposure Assessment of Nickel in Zhejiang Province, China.pdf). Page was originally created 2026-05-14 by autonomy_phase1_v0; this pass enhances per the v2 manual-fetch skill. Changes:
  • Body-weight table previously carried a misattributed annotation (“n=4173 of 19,000 (16.60%) population subset” against 0–6 years; “70.14% of survey” against 18–59; “13.26% of survey” against ≥60). Per p. 4 of the source, the 16.60% / 70.14% / 13.26% shares describe the three-bucket population census grouping (0–14, 15–59, ≥60), not the five-bucket consumption-survey grouping used in Tables 2–4. The misattributed annotations were removed and a correctly attributed sentence about the three-bucket grouping added below the body-weight table.
  • Methods section rewritten to ”## Methods (brief)” per current Part 6 template, with instrument vendor/model, digestion vendor/model, reagent vendor, tuning solution vendor, USEPA EDI/HQ/THQ equation forms, and Mode A–D definitions made explicit. Vendor/material names retained per Part 12 Exception 2 (scientific-method vendor names are not brand-firewall violations).
  • Implications section softened: removed prescriptive language (“HMTc Ni thresholds … should anchor on these distributions”) in favour of contribution-framed wording (“contributes occurrence and exposure data … relevant to Cat 1 modelling”) per Part 2 wiki/HMTc firewall.
  • Sample-size discrepancy between paper text (Section 3.1) and Table 1 was preserved with both totals shown, since the source itself is internally inconsistent (n=1404 vs 1405 aquatic; n=691 vs 690 meat; n=392 vs 365 vegetables; n=145 vs 136 fruits). Key numbers table uses Table 1 values throughout (the more granular source).
  • products: ["[[products/processed-meats]]"] added — sausages (n=316), Chinese bacon (n=296), and “other meat product” (n=7) all map to the existing wiki/products/processed-meats.md page; the paper’s “Meat and products” category is processed meat. Canned-fish product already declared.
  • Frontmatter sampling_locations populated and sampling_year_range filled in. raw_handle, cite_key, raw_path, license, doi, near_duplicates preserved.
• Source-internal inconsistency on age-bucket labelling: p. 5 of the paper says respondents were classified into five groups “0–6, 6–13, 11–17, 18–59, and ≥60 years” (overlapping at 6 and 11), but Tables 2, 3, and 4 use non-overlapping groups 0–6, 7–10, 11–17, 18–59, ≥60. The wiki follows the table groupings since all numerical attributions are tied to the tables.
• 2026-05-29 (audit application): Fresh-context Agent subagent (general-purpose) ran the five-check audit per docs/gpt-collaboration/audit-prompt.md. Verdict: PROMOTE with 0 ❌ findings and 3 ⚠️ framing concerns. Concerns applied to this commit: (a) §2.1-text vs §3.1/Table 1 sample-size discrepancy framing made explicit (prior wording “paper text; Table 1 totals” was imprecise since both totals are paper-text); (b) “Aquatic: Other processed” row label expanded to “Other processed aquatic products” to match Table 1; (c) EFSA TDI chronology phrasing rewritten to match the paper’s “initially set 2.8 in 2015 → later modified to 13” order. No false-positive findings to record. Routing audit re-run 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.