BfR 2022 — Nickel dietary intake via food based on the BfR MEAL Study (Germany)

The German Federal Institute for Risk Assessment (BfR) conducted a total diet study (TDS) on nickel in food as part of the first German MEAL Study, analyzing 356 foods grouped into 19 main food categories and measuring Ni concentrations in 840 pools purchased nationwide between December 2016 and May 2019. The study found that the food groups “Legumes, nuts, oil seeds and spices” and “Coffee, cocoa and tea” carry the highest mean nickel concentrations at 1,583 µg/kg and 1,488 µg/kg respectively, with cocoa powder the single highest item at 11,050 µg/kg, followed by cashew nuts at 5,350 µg/kg. Dietary nickel intake for German adults and adolescents averages 11% of the EFSA-derived TDI of 13 µg/kg BW/day (P50, UB), while children average 42% of the TDI (P50, UB), with fewer than 5% of highly exposed children (ages 0.5–5 years) exceeding the TDI. This report provides the most geographically specific German-population nickel exposure baseline, lower than EFSA’s pan-European estimates for the same population when using German consumption data paired with pan-European occurrence data.

Key numbers

Nickel concentrations by major food group (mean, upper bound, µg/kg):

• Grains and grain-based products: mean 359 µg/kg, range min 20 – max 1,975 (chia seeds highest within group)
• Vegetables and vegetable products: mean 84 µg/kg, range min 17 – max 340 (vegetable crisps)
• Starchy roots/tubers: mean 50 µg/kg, range min 23 – max 100 (fried potatoes/potato chips, UB)
• Legumes, nuts, oil seeds and spices: mean 1,583 µg/kg, range min 60 – max 5,350 (cashew nuts)
• Fruit and fruit products: mean 112 µg/kg, range min 16 – max 1,080 (dried fruits)
• Meat and meat products: mean 27 µg/kg, range min 16 – max 60 (liver sausage, poultry)
• Fish and seafood: mean 30 µg/kg, range min 16 – max 165 (shellfish)
• Milk and dairy products: mean 48 µg/kg, range min 6 – max 295 (ice cream)
• Eggs and egg products: mean 17 µg/kg
• Sugar, confectionery and water-based sweet desserts: mean 601 µg/kg, range min 30 – max 2,800 (semisweet/dark chocolate)
• Animal and vegetable fats and oils: mean 65 µg/kg, range min 30 – max 100 (corn/sunflower oil); all samples <LOQ (100 µg/kg)
• Fruit and vegetable juices and nectars: mean 21 µg/kg, range min 20 – max 27 (multi-vitamin fruit juice)
• Water and water-based beverages: mean 18 µg/kg, range min 3 – max 44 (energy drinks)
• Coffee, cocoa, tea and infusions: mean 1,488 µg/kg, range min 13 – max 11,050 (cocoa powder)
• Alcoholic beverages: mean 19 µg/kg, range min 6 – max 48 (red wine mLB / spirits UB)
• Food products for infants and toddlers: mean 185 µg/kg, range min 20 – max 1,040 (porridge/millet powder)
• Vegan/vegetarian products: mean 386 µg/kg, range min 48 – max 1,000 (soy protein extrudate)
• Composite dishes: mean 53 µg/kg, range min 20 – max 220 (lentil/pea/bean soup)
• Sauces and condiments: mean 111 µg/kg, range min 10 – max 565 (soy sauce)

Top 10 individual foods by nickel concentration (µg/kg, UB):

1. Cocoa powder: 11,050
2. Cashew nuts: 5,350
3. Sunflower seeds: 4,750
4. Walnuts: 4,300
5. Semisweet/dark chocolate: 2,800
6. Beverage powder (total): 2,150
7. Hazelnuts: 2,000
8. Hazelnut spread: 2,000
9. Chia seeds: 1,975
10. Pumpkin seeds: 1,850

Dietary exposure estimates (Tables 3 and 4):

• Adults and adolescents (N=13,926): mean UB 1.6 µg/kg BW/day; P50 1.4 µg/kg BW/day; P95 3.2 µg/kg BW/day. TDI exhaustion: mean 12%, P50 11%, P95 24%.
• Children (N=732, ages 6 months–<5 years): mean UB 6.1 µg/kg BW/day; P50 5.5 µg/kg BW/day; P95 10.6 µg/kg BW/day. TDI exhaustion: mean 47%, P50 42%, P95 82%.
• Children aged 1–<2 years exhibit the highest exposure: mean UB 6.3 µg/kg BW/day; P50 5.8 µg/kg BW/day; P95 11.0 µg/kg BW/day (TDI exhaustion mean 48%, P50 45%, P95 84%).
• Children aged 6 months–<1 year (the lowest-exposure pediatric subgroup): mean UB 4.9 µg/kg BW/day; P50 4.4 µg/kg BW/day; P95 9.0 µg/kg BW/day (TDI exhaustion mean 38%, P50 34%, P95 69%).
• Highly exposed children exceeding TDI: 2% of children ages 0.5–5 years overall (n=15 children exceeding the TDI out of N=732 in the cohort); 3% in the 1–<2 years subgroup; 1% in the 6 months–<1 year subgroup. EFSA 2020 framing: <5% of high-consumer children exceed the TDI in this MEAL assessment, vs typically exceeding the TDI in the EFSA estimate for the same age groups.

Largest contributors to adult nickel exposure (% of total, UB; Figure 1): Grains and grain-based products 24%; Coffee, cocoa and tea 20%; Water and water-based beverages 11%; Composite dishes 7%; Sugar/confectionery/sweet desserts 7%; Fruit and fruit products 6%; Milk and dairy products 5%; all other groups <5% each (total “Other” 20%).

Largest contributors to children’s nickel exposure (% of total, UB; Figure 2): Grains and grain-based products 28%; Sugar/confectionery/sweet desserts 12%; Milk and dairy products 10%; Food products for infants and toddlers 9%; Composite dishes 8%; Fruit and fruit products 8%; Coffee/cocoa/tea 7%; Water and water-based beverages 5%; all other groups <5% each (total “Other” 14%).

High-consumer exposure to specific foods — adults and adolescents (Table 7, UB µg/kg BW/day; consumers only, P50): Cashew nuts (n=80, 0.6% of population) 3.52; Cocoa powder (n=77, 0.6%) 3.10; Soy protein extrudate (n=20, 0.1%) 3.08; Trail mix (n=66, 0.5%) 3.06; Soy milk (n=103, 0.7%) 3.05; Walnuts (n=403, 2.9%) 2.67; Dried fruits (n=193, 1.4%) 2.61; Porridge (n=19, 0.1%) 2.80; Lentils (n=46, 0.3%) 1.97; Hazelnut spread (n=148, 1.1%) 1.84. All foods have consumer fractions <10% of the adult population (rare-consumption pattern).

High-consumer exposure to specific foods — children (Table 8, UB µg/kg BW/day; consumers only, P50): Soy milk (n=7, 1.0%) 11.95; Porridge (n=18, 2.5%) 9.02; Cocoa powder (n=38, 5.2%) 8.76; Trail mix (n=4, 0.5%) 8.44; Millet (n=14, 1.9%) 7.68; Oatmeal (n=80, 11%) 7.40; Walnuts (n=24, 3.3%) 7.14; Spelt bread (n=11, 1.5%) 6.93; Sunflower seeds (n=10, 1.4%) 6.71; Cashew nuts (n=6, 0.8%) 6.59. With the exception of oatmeal (~11% of children consume it), all are rarely-consumed foods.

European comparison (Tables 11 and 12, mean UB µg/kg BW/day): Germany (MEAL): 1.3–1.6 (mLB–UB) for adults 15–80; 5.4–6.1 (mLB–UB) for children 0.5–<5 years. Italy (Cubadda 2020): 1.47–1.55 (adults), 4.0 (children <3). France (Arnich 2012): 2.33 (adults), 3.83 (children 3–17). Spain Canary Islands (Gonzalez-Weller 2012): 1.4 (adults 18–75). United Kingdom (Rose 2010): 1.49–1.63 (adults LB–UB), 4.17–4.87 (children 1.5–4.5 LB–UB). EFSA (2020) estimate using EFSA pan-European occurrence + German consumption: 3.34–3.93 (adults 18–<65 LB–UB), 8.24–9.91 (children 1–<3 LB–UB) — roughly 2× the MEAL estimate, attributable to differences in occurrence-data sourcing.

EFSA TDI reference point: 13 µg/kg BW/day, derived from BMDL10 of 1.3 mg Ni/kg BW/day from rat developmental toxicity studies (one- and two-generation studies, post-implantation loss as the critical effect), with default uncertainty factor of 100 (EFSA 2020). This replaces the prior 2.8 µg/kg BW/day value; the update reflects revised BMD methodology guidance (EFSA 2017). For acute sensitization-related effects in nickel-allergic individuals, EFSA derived a LOAEL of 4.3 µg/kg BW for the MoE approach (EFSA 2020); the BfR MEAL Study is methodologically oriented toward long-term exposure only and cannot inform the acute-effect assessment.

Methods (brief)

Total diet study design: 356 foods representing >90% of German food consumption, purchased in four regions nationwide (national, east, south, west, north) between December 2016 and May 2019 across non-seasonal, season-1, and season-2 purchasing windows, and across non-specific, organic, and conventional production types. Foods prepared as ready-to-eat per typical German household practice, then pooled (15–20 individual items per pool) and homogenized. Of the 356 foods, 105 were sampled with stratification by production type and 70 by region; the remainder were non-stratified. Nickel was analyzed in 840 total pools; 23% (193 of 840) showed non-quantifiable nickel concentrations, with the highest non-detect proportions in “Animal and vegetable fats and oils” (100%) and “Eggs and egg products” (90%). Limits of detection and quantification varied by food category (drinking water LOD 0.3 µg/kg, LOQ 1 µg/kg; most other foods LOD 6–30 µg/kg, LOQ 20–100 µg/kg). Results were calculated using both the modified lower bound (mLB; values <LOD set to 0, values between LOD and LOQ set to LOD) and upper bound (UB; values <LOD set to LOD, values <LOQ set to LOQ) approaches. Differences between organic and conventional production were minor, with no consistent tendency toward higher concentrations in either, so the published occurrence and exposure assessment is based on conventional and non-specified-production pools only.

Consumption data: adults and adolescents from NVS II (Nationale Verzehrsstudie II; 2005–2006; n≈20,000 surveyed aged 14–80, n=13,926 with both 24-hour recalls used), conducted by the Max Rubner Institute (Krems et al. 2006; MRI 2008). Children from the VELS study (Verzehrsstudie zur Ermittlung der Lebensmittelaufnahme von Säuglingen und Kleinkindern; 2001–2002; n=816 infants and toddlers 6 months to <5 years, n=732 non-breastfed used for the exposure assessment; Heseker et al. 2003; Banasiak et al. 2005), based on 3-day weighing protocols. Exposure was estimated under both mLB and UB; only UB values are summarized in the main text since mLB–UB differences are described as minor by the authors.

Key limitations stated by the authors: consumption data from 2005–2006 (adults) and 2001–2002 (children) may not reflect current dietary patterns (the more recent KiESEL study for children was not yet available at the time of evaluation); the 24-hour recall and 3-day weighing methods may underestimate intake from rarely consumed high-nickel foods; the MEAL drinking water used for meal preparation had a relatively high Ni concentration at 3 µg/kg (mean UB) vs 1 µg/kg in regionally sampled MEAL drinking water (n=29), which may overestimate nickel exposure from water-prepared foods such as tea, coffee, and infant formula in regions with lower-Ni tap water. The MEAL food list covers >90% but not 100% of consumed foods, so total exposure may be modestly underestimated. Other potential exposure routes (tobacco smoke, dust/soil) are not included.

Implications

Certification: Nickel is in the HMT&C analyte vocabulary. This report establishes that cocoa powder (11,050 µg/kg), cashews (5,350 µg/kg), sunflower seeds (4,750 µg/kg), walnuts (4,300 µg/kg), and dark chocolate (2,800 µg/kg) are the dominant nickel-concentrated foods in the German food supply. Any HMT&C product category containing these ingredients will have nickel as a primary concern. The high-consumer rare-food pattern (Tables 7 and 8: cocoa powder, cashews, soy milk, soy protein extrudate, oatmeal, trail mix, porridge, walnuts) is the dominant exposure driver at the population tail and is relevant to product categories that contain those ingredients as the primary base.

Courses: The BfR MEAL Study is the definitive German total diet study for nickel and provides the concentration data for estimating nickel exposure from processed foods and snacks. Children’s exposure substantially exceeds adults’ on a body-weight basis (mean UB 6.1 vs 1.6 µg/kg BW/day), and grains are the dominant pathway at population scale even though coffee, cocoa, and nuts carry higher concentrations per kg of food. The roughly 2× gap between the MEAL estimate and the EFSA (2020) estimate for Germany using the same consumption data is a useful teaching example for how occurrence-data sourcing (national TDS vs pooled European data) affects exposure estimates.

App: Cocoa powder, cashews, sunflower seeds, walnuts, hazelnut spread, dark chocolate, soy milk, soy protein extrudate, oatmeal, and porridge/millet should carry high nickel flags in the ingredient risk model. Soy-based products (soy milk, soy protein extrudate) are high-exposure drivers for the small fraction of the population consuming them frequently. Infant formula prepared with high-nickel tap water can be an additional exposure pathway via the drinking-water route per the paper’s uncertainty discussion, though the MEAL Study itself does not measure infant formula products directly — the “Food products for infants and toddlers” food group in this study covers weaning porridges, millet powders, and similar complementary foods rather than infant formula proper.

Microbiome: Not addressed in this report.

Verification notes

Merge-enhance pass 2026-05-20 (Claude Opus 4.7, autonomous ingest cycle; existing page updated: 2026-05-13 predated the 2026-05-14 schema cutoff and the v2 audit workflow):

• Corrected transposed Children-total UB values. Prior page reported children total (N=732) as mean UB 6.2 and P95 UB 10.8 µg/kg BW/day. Verified against PDF Table 4 (p. 10): total UB mean is 6.1 and P95 is 10.6; the 6.2/10.8 figures are the Boys subgroup (N=368) UB mean and P95, not the total. Corrected to 6.1 and 10.6 in Key numbers. P50 5.5 was correct (matches both total and boys at the same value).
• Added Water and water-based beverages 5% to children’s contributors line. Prior list stopped at “Coffee/cocoa/tea 7%”; the Figure 2 pie chart (p. 11) also shows Water 5% as a named slice before the “Other (<5%)” 14% aggregate. Added for completeness so the named slices sum cleanly (28+12+10+9+8+8+7+5 = 87% plus 14% Other = 101%, consistent with rounding).
• Added Tables 7 and 8 high-consumer per-food anchors. These are load-bearing for downstream synthesis: they identify which specific foods drive the upper-tail exposure in adults (cashews, cocoa powder, soy protein extrudate, trail mix, soy milk, walnuts, dried fruits, lentils, hazelnut spread, porridge) and children (soy milk, porridge, cocoa powder, trail mix, millet, oatmeal, walnuts, spelt bread, sunflower seeds, cashews), with consumer fractions for each. Without these, the page captured only food-group contributions but missed the food-specific tail.
• Added 6 months–<1 year and 1–<2 years children subgroup detail. The 1–<2 year subgroup has the highest pediatric exposure (mean UB 6.3, P95 11.0 µg/kg BW/day; TDI exhaustion P95 84%); the 6 months–<1 year subgroup has the lowest (mean UB 4.9, P95 9.0; TDI exhaustion P95 69%). These were partially in the prior page (1–<2 years P50 and P95 only) but not the means, TDI exhaustion percentages, or the 6 months–<1 year subgroup. Useful for synthesis of pediatric nickel exposure patterns.
• Added European comparison anchors (Tables 11 and 12). The prior page had a single summary sentence on Germany-vs-EFSA divergence; the merged version preserves that summary and adds the specific values for Italy, France, Spain (Canary Islands), and the United Kingdom (Tables 11 and 12), since these are useful triangulation points for European nickel exposure synthesis.
• Expanded Methods detail. Added explicit description of the four regional sampling windows (national/east/south/west/north), the two seasonal sampling windows, the three production-type sampling tracks, the non-quantifiable-pool percentages by food group (100% for fats/oils, 90% for eggs), and the explicit acknowledgment by the authors that the KiESEL study (more recent children’s consumption data) was not yet available at the time of evaluation.
• Clarified TDI hazard-characterisation framing. The TDI of 13 µg/kg BW/day replaces the prior 2.8 µg/kg BW/day value; the update reflects revised BMD methodology guidance (EFSA 2017). Also surfaced the LOAEL of 4.3 µg/kg BW for acute sensitization-related effects in nickel-allergic individuals and the methodological note that the MEAL Study cannot inform acute-effect assessment (long-term exposure only).
• Labeled the lead-paragraph “11% / 42% of TDI” framing as P50, UB. The prior page paraphrased the BfR summary directly; the merged version preserves the same numbers but adds the (P50, UB) qualifier so the reader does not confuse them with the means (12% and 47% respectively) reported in Tables 9 and 10.
• Corrected legacy slug in ## Wiki pages updated on ingest. Prior entry [[ingredients/oats]] had no corresponding page; wiki/ingredients/oat.md is the canonical slug (matching the frontmatter declaration). Corrected to [[ingredients/oat]]. The mismatch was the result of pre-2026-05-14 ingest writing both slug forms inconsistently.
• Preserved cite_key, raw_handle (manual-fetch-kimi per the existing-page convention), raw_path, license, evidence_tier, doi (null with no_doi_assigned: true), access_url, sample_n, sample_population, jurisdictions, source_type, publication, matrices array. No new ingredient, product, regulation, or metal pages proposed. No brand-by-brand listings appear in the source; Part 12 firewall posture unchanged.

Phase 2 audit application 2026-05-20 (fresh-context Agent subagent, verdict REVISE):

• Applied ❌ slug-vocabulary finding: [[ingredients/dark-chocolate]]. Auditor flagged that no wiki/ingredients/dark-chocolate.md exists in the current taxonomy snapshot; the canonical ingredient slug for dark and semisweet chocolate is [[ingredients/chocolate]]. Verified independently: ls wiki/ingredients/dark-chocolate.md errors with “No such file or directory”; wiki/ingredients/chocolate.md is present; the taxonomy snapshot ingredients list contains chocolate and cocoa but no dark-chocolate. Corrected the frontmatter ingredients array entry and the Wiki-pages-updated list entry (the legacy [[ingredients/dark-chocolate]] was a pre-2026-05-14 invented sub-variant slug that the merge-enhance carried forward unchecked). The 2,800 µg/kg “semisweet chocolate/dark chocolate” Table 1 / Table 2 value is preserved verbatim in Key numbers prose; only the slug routing target changed.
• Applied ❌ slug-vocabulary finding: [[products/infant-formula]]. Auditor flagged two issues: (a) no bare products/infant-formula page exists in the taxonomy (only sub-variant pages: powder-non-soy, powder-soy-based, rtf-liquid-non-soy, rtf-liquid-soy-based, etc.), and (b) the MEAL Study does not measure infant formula products directly — the only mention in the paper is in the uncertainty discussion about drinking-water Ni potentially affecting infant formula prepared with tap water. Verified independently: ls wiki/products/infant-formula.md errors; the routing audit was carrying the bare infant-formula slug as a broadcast key to the four sub-variant pages via broad_formula_context, but this routing is not supported by direct measurement in the paper. Removed [[products/infant-formula]] from the frontmatter products array. Downstream routing change: the paper will no longer fan out to the four infant-formula sub-pages as broad context. This is the correct outcome because the paper genuinely does not provide infant-formula context — the “Food products for infants and toddlers” food group covers weaning porridges and millet powders, not infant formula. The App-block in Implications is updated to clarify this distinction.
• Applied ⚠️ slug-vocabulary finding: [[regulations/efsa-ni-contam-2020]]. Auditor flagged that this slug is not in the regulations taxonomy snapshot; the canonical regulation slug for the EFSA 2020 nickel TDI of 13 µg/kg BW/day is [[regulations/efsa-nickel-tdi]]. Verified independently: ls wiki/regulations/efsa-ni-contam-2020.md errors; wiki/regulations/efsa-nickel-tdi.md is present; the taxonomy snapshot regulations list contains efsa-nickel-tdi but no efsa-ni-contam-2020. Corrected the Wiki-pages-updated list entry.
• Applied ⚠️ Check-1 N=15 parenthetical clarification. Auditor flagged that the prior wording “2% of children ages 0.5–5 years overall (N=15)” could mislead a casual reader into thinking N=15 is sample size rather than the count of children exceeding the TDI. PDF p. 17 confirms N=15 is the numerator (children exceeding TDI), not the cohort N. Clarified to “n=15 children exceeding the TDI out of N=732 in the cohort” to remove the ambiguity.
• Did not change the numerical fidelity content of Key numbers (auditor confirmed ✅ on all verified table cells, exposure values, TDI exhaustion percentages, food group means, Table 2/7/8 high-concentration and high-consumer foods, methods particulars), the Part 12 brand firewall posture (no brand names anywhere; trivially clean for a government TDS), or the Part 2 wiki/HMTc firewall posture (no threshold proposals, no softening/strengthening, P50/UB framing correctly labeled in lead paragraph). Auditor noted Sirot 2018 omission from the France-children comparison as optional polish; declined to add given that the prior page captured the more comparable Arnich 2012 entry and the synthesis pass can add additional triangulation points if needed (scope creep relative to the audit-application pass).
• No findings rejected as false positives. All four applied findings were verified against the actual file system / taxonomy snapshot / PDF before applying.

Wiki pages updated on ingest

• nickel
• cocoa
• cashew-nuts
• chocolate
• oat
• wheat
• soy
• walnuts
• sunflower-seeds
• snacks-crackers-biscuits
• chocolate
• efsa-nickel-tdi

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.