BfR MEAL Study Final Report — Germany’s first Total Diet Study (2023)

This is the umbrella final report of the BfR MEAL Study (Mahlzeiten für die Expositionsschätzung und Analytik von Lebensmitteln — “meals for exposure assessment and analysis of foods”), Germany’s first Total Diet Study (TDS), published as a BfR Science Report on 28 July 2023. The report documents the design, implementation, quality assurance, and use of the MEAL data set; analyte-specific concentration tables and exposure assessments are published in companion peer-reviewed articles and BfR opinions/communications, several of which already have wiki source pages (Fechner et al. 2022 for Hg, Cd, Pb, Ni; Sarvan et al. 2021 for MeHg in fish; Hackethal et al. 2021 for inorganic arsenic and arsenic species; BfR Communication 033/2022 for the Ni intake assessment; BfR Opinion 023/2024 for the MeHg fish risk assessment). The final report is the methodology and programmatic source-of-record for that data set; for HMI’s purposes, it documents the sample basis, analytical scope, and limit-of-quantification regime that the companion publications draw from, and it lists the EU maximum-level exceedances identified across the entire study.

The report is distinct from the four MEAL-derived sources already in the corpus: fechner2022-bfr-meal-hg-cd-pb-ni-germany.md (Hg, Cd, Pb, Ni paper), bfr2022-nickel-dietary-intake-germany-meal.md (Ni communication 033/2022), bfr2024-methylmercury-fish-seafood-germany.md (MeHg fish opinion), and hopfner2025-infant-formula-dietary-exposure-germany.md (infant formula exposure paper). All four reuse occurrence data documented in this final report.

Key numbers

Scope and sample basis.

• 356 foods on the core-module food list, selected to cover >90% of food and beverage consumption in each of six age strata using NVS II (14–80 yr; n=13,926 / 14,484 over 2005–2006) and VELS (<1–4 yr; n=804 over 2001–2002) consumption data, supplemented with foods of historic concern or emerging dietary relevance (e.g., chia seeds, striped catfish, avocado, meat substitutes).
• 869 pooled samples for the core module (868 for nitrate), comprised of 13,552 subsamples (13,532 for nitrate); for 151 of the 356 foods, additional pooled samples were prepared to capture region (four regions: North, East, South, West), season (two seasons of varying length), or production type (conventional vs organic; 105 foods received separate organic pools).
• Approximately 56,750 single food items purchased over 4 years 8 months across 149,499 km of nationwide procurement (introduction summary; the report’s section 4.2 separately gives a rounded ~60,000 figure for gross purchasing volume).
• 336 individual substances analysed in 9 modules: core (elements and environmental contaminants), perfluoroalkyl substances, mycotoxins, process contaminants, nutrients, substances migrating from food contact materials, pesticide residues, pharmacologically active substances, and food additives.
• 36 analytical methods executed across one internal laboratory, eight commercial laboratories, two other institutions, and one state investigation office. >140,000 individual analytical results returned.

Core-module element substance list (HMI-relevant). Antimony (Sb), aluminium (Al), arsenic (total arsenic and the speciated set: inorganic arsenic [iAs], arsenobetaine [AsB], dimethylarsinic acid [DMA], monomethylarsonic acid [MMA]), barium, lead (Pb), cadmium (Cd), cobalt, lithium, nickel (Ni), mercury (tHg, with methyl-mercury [MeHg] additionally analysed in fish/seafood and mushroom-based dishes), silver, thallium, vanadium, and tin (Sn). Chromium and uranium are not in the MEAL substance list.

Sample structure of the heavy-metal-relevant sub-modules.

• Elements + nitrate (Fechner et al. 2022): 869 pooled samples / 13,552 subsamples / 356 foods across all 19 main food groups (data in brackets in Table 4 refers to nitrate: 868/13,532/355).
• Inorganic arsenic and arsenic species (Hackethal et al. 2021): 73 pooled samples / 1,411 subsamples / 55 foods, concentrated in fish/seafood (39/720/30) plus rice and rice-based products, with 19 additional pools selected for previously-elevated total arsenic (e.g., algae, boletus, bovine liver).
• Methyl mercury (Sarvan et al. 2021): 49 pooled samples / 871 subsamples / 34 foods, concentrated in fish/seafood (39/720/30) and extended to mushrooms (3 foods/91 subsamples) and one composite-dish food on expert-group recommendation.
• Organotin compounds (seven compounds including TBT, DBT, MBT, TPhT, DPhT, MPhT): 39 pooled samples / 720 subsamples / 30 foods, fish/seafood/invertebrates only.

EU maximum-level / maximum-residue-level exceedances reported to BMEL (Table 38). No exceedances for the HMI-tracked heavy metals (Pb, Cd, tAs, iAs, tHg, MeHg, Ni, Al, Sn, Sb) appear in this table — heavy-metal levels across the 356 core-module foods were below applicable EU maximum levels at the pool level. Reported exceedances for other substance groups (preserved here for cross-module context): non-dioxin-like PCB (ICES-6 sum) in spiny dogfish 190 ng/g against a 200 ng/g maximum level (note: this value sits just under the limit; the table lists it as a notification-relevant finding rather than a confirmed exceedance); copper exceedances of EU feed/food limits in chia seeds (16.5 mg/kg vs 10 mg/kg), beef liver (66.4–119 mg/kg vs 30 mg/kg), sheep’s liver (68.1–77.6 mg/kg vs 30 mg/kg), red deer / roe deer (2.35 mg/kg vs 0.01 mg/kg), wild boar (2.03 mg/kg vs 0.01 mg/kg), and honey (0.265–0.355 mg/kg vs 0.01 mg/kg); aflatoxin B1 exceedances in buckwheat (2.66 µg/kg vs 2 µg/kg ML) and chia seeds (4.22 µg/kg vs 2 µg/kg ML), plus a notification-relevant near-miss in pistachios (7.53 µg/kg against an 8 µg/kg ML); ochratoxin A in buckwheat (5.86 µg/kg vs 3 µg/kg ML); and chlorate exceedances of the 10 µg/kg MRL in 11 processed foods ranging 19–105 µg/kg. Copper is outside the HMI metals scope; the chlorate/mycotoxin/PCB findings are recorded here only to anchor the report’s exceedance methodology.

Limit-of-quantification regime. The analytical tender specified the lowest LOQs that could be reliably maintained across every food matrix in scope. Selection criterion: for each substance, mean exposure should remain below the relevant health-based guidance value even when 100% of the content data was left-censored at the LOQ. Higher LOQs were accepted only when closing existing data gaps was deemed essential and lower LOQs were analytically infeasible. The companion publications (e.g., Fechner et al. 2022) report mercury LOQ = 0.002 mg/kg moist / 0.005 mg/kg dry, cadmium LOQ = 0.001 mg/kg moist, lead LOQ = 0.002 mg/kg moist, nickel LOQ = 0.02 mg/kg moist.

Companion publication inventory (Table 39, last updated by the report in 2023). Seven occurrence-data papers (Stadion et al. 2023 corrigendum; Schendel et al. 2022 retinol/β-carotene; Stadion et al. 2022 dioxins/dl-PCBs; Fechner et al. 2022 Hg/Cd/Pb/Ni; Schwerbel et al. 2022 Ca/K/P; Hackethal et al. 2021 total/water-soluble arsenic species; Ptok et al. 2020 Cd/Pb), three exposure-assessment papers (Kolbaum et al. 2023 copper; Hackethal et al. 2023 chronic arsenic exposure; Sarvan et al. 2021 methyl-mercury exposure), nine methodology/communication papers, and six BfR opinions and communications (033/2022 nickel; 037/2018 ndl-PCBs; 005/2021 and 026/2022 iodine; 006/2023 sweeteners; plus the 2022 BMUV ndl/dl-PCB decree).

Methods

Total Diet Study design (TDS). The study followed the EFSA/FAO/WHO 2011 harmonised TDS guidance (EFSA Journal 9(11):2450) further developed in the EU “TDS Exposure” project (Kolbaum et al. 2019). Three core TDS principles applied: (1) coverage of the most-consumed foods to represent average consumption; (2) preparation of foods as consumers would prepare them at home, including effects of processing; (3) pooling of similar foods for cost-effective analysis. The methodology adapts to Germany the approach used in TDS implementations in France, Portugal, the Czech Republic, Korea, the USA, and New Zealand (more than 50 countries worldwide as of 2023).

Food-list construction. Foods were drawn from NVS II (German National Nutrition Survey II) for adolescents and adults and from the VELS study (Verzehrsstudie zur Ermittlung der Lebensmittelaufnahme von Säuglingen und Kleinkindern für die Abschätzung eines akuten Toxizitätsrisikos durch Rückstände von Pflanzenschutzmitteln) for infants and small children (<1–4 years). Foods covering 90% of consumption in each age stratum were selected; additional foods were added for historic-concern substances and emerging dietary trends. The 2013 primary concept estimated 350 foods; the final list grew to 356 after expert-group consultation.

Sampling strata. Germany was subdivided into four regions (North, East, South, West) rather than the originally planned six, after the international advisory board reviewed climate and soil-contamination data showing limited regional variation. In each region, three sample points were selected using the BIK community-size classification: one large city (>100,000 inhabitants), one medium-sized city (20,000–100,000), and one rural area (<20,000). Subsample counts at each point were weighted by the BIK community-size population share within the region (e.g., a 15-subsample East regional pool drew 9 subsamples from the large city, 4 from the medium-sized city, and 2 from the rural area). Subsamples per pool: 20 where no further stratification, 15 where additional region/season/production-type pools were prepared. Stratified four-region pools consisted of 4×15 = 60 subsamples; two-season pools 2×15 = 30 subsamples. Tap water was sampled separately at 30 nationwide sites (29 publicly accessible taps plus the MEAL study kitchen).

Food preparation. A purpose-built study kitchen at Berlin Alt-Marienfelde, equipped with standard household-grade appliances (Tarrington House MWD5130 microwave; MKN CVEKOI2 induction cooker; MKN Master of Performance electric oven; MKN SpaceCombi MagicPilot combi steamer; Meiko DV 80.2 pass-through dishwasher), prepared each subsample according to a documented recipe plan based on representative consumer-preparation studies (telephone household survey, n=1,008, and online survey, n=2,003, both fielded by aproxima Gesellschaft für Markt- und Sozialforschung mbH). In field phase 1, 64% of subsamples were cooked in the study kitchen, 10% were washed and chopped only, and 26% were purchased ready-to-eat; in field phase 2 the corresponding figures were 53%, 12%, and 35%.

Homogenisation. Two cutting mills (Retsch GM200 and GM300) with food-contact components selected per analyte class: polypropylene container plus titanium-niobium-coated or Eifeler Carbon X-coated stainless-steel blades for element analysis (to avoid Ni leaching), stainless steel with amber-glass storage for lipophilic analytes, and cauterised glass plus aluminium foil for food-contact-material migration analytes. Liquid nitrogen or dry ice was used for thermolabile analytes and to inactivate enzymes that could alter analyte levels. Ultrapure water from a Merck Milli-Q Integral 5 system. Mycotoxin subsamples were homogenised separately to avoid cross-contamination.

Analytics. 36 analytical methods deployed across 12 laboratories (1 internal, 8 commercial, 2 other institutions, 1 state investigation office). Element analyses (Fechner et al. 2022) used ICP-MS for Cd/Pb/Ni and direct mercury analysis for tHg; arsenic speciation (Hackethal et al. 2021) used HPLC-ICP-MS; PFAS used HPLC-HRMS; food additives used various LC-MS/MS multimethods. Quality assurance covered all stages under DIN EN ISO 9001 (Table 2): barcode-based traceability of every food item from purchase to homogenate, temperature-logged storage and transport (EXTECH SD200), interlaboratory comparisons, accreditation requirements, reference-material analysis, and 5–10% blind QA re-analysis per analyte. A 20–30% plausibility check on all analyte results compared MEAL values against 10 years of national food-monitoring data (BfR ZEBS), WHO GEMS/Food, EFSA occurrence opinions, and other scientific literature.

Left-censored data. Generally reported under upper-bound (UB) and modified lower-bound (mLB) approaches in the companion papers; the final report does not re-publish concentration tables.

Implications

Certification. This umbrella final report does not itself report per-food concentration values for HMI-tracked metals; the values used for downstream HMTc work are in the companion peer-reviewed papers (Fechner 2022, Sarvan 2021, Hackethal 2021, Ptok 2020). The report’s contribution to the certification programme is methodological: it documents the sample basis, food-list construction, region/season/production-type stratification, and ISO 9001 quality assurance underlying the German-market occurrence data that several HMI source pages already use. The Table 38 exceedance summary anchors the substantive finding that all heavy-metal results at the pool level were below applicable EU maximum levels for the HMI-tracked metals over the 2017–2019 field phase.

Courses. The report is a teaching case for Total Diet Study methodology in a regulator-grade implementation: the modular study design, the pool-vs-subsample structure, the matrix-aware LOQ selection rule, the homogenisation tooling choices that prevent analyte introduction (titanium-coated cutting blades for elements; cauterised glass and aluminium foil for migrants), and the 20–30% plausibility-check regime against national monitoring and international databases. It is also a worked example of the EFSA/FAO/WHO 2011 harmonised TDS guidance applied at national scale.

App. The downstream MEAL public-use file (https://www.bfr-meal-studie.de/en/public-use-file-en.html) provides German-market food-by-food concentration data for the analytes listed above; the present source page documents the public-use file’s provenance, sample basis, and limit-of-quantification regime so that downstream consumers of the public-use file can cite the umbrella study correctly.

Microbiome. Not directly addressed.

Wiki pages this source may touch

• lead
• cadmium
• arsenic
• arsenic-inorganic
• arsenic-total
• mercury
• mercury-total
• mercury-methyl
• nickel
• aluminum
• tin
• antimony

Verification notes

• Fresh ingest 2026-06-02 from raw/manual-fetch/Kimi_Agent_Download Corruption Issue/condiments2_papers/04_Chocolate_Seasonings_Salt/06_BfR_MEAL_Study_Final_Report.pdf via the v2.0 manual-fetch skill. Identity checks (DOI/raw_handle/cite-key) returned no matching wiki page; the four MEAL-derived sources already in the corpus (Fechner 2022, BfR 2022 Ni, BfR 2024 MeHg, Höpfner 2025 infant-formula) are downstream products of the data set this report documents, not duplicates of the report itself. Author field uses the institutional author (BfR) because the science report carries no individual author byline. DOI: null — the BfR Science Report does not carry a DOI; access_url points to the BfR MEAL Study homepage. License recorded as “BfR public report” (German federal agency publication, free public access). Sample structure figures captured verbatim from Tables 4–8 (core elements 869/13,552/356; arsenic species 73/1,411/55; methyl mercury 49/871/34; organotin 39/720/30); the bracketed nitrate-only counts (868/13,532/355) are footnoted in Table 4. Frontmatter ingredients/products are intentionally empty because the report’s scope is the entire 356-food list, not a single ingredient or product; routing should flow through the companion publications that name specific foods. Methods preserves manufacturer/model names for analytical reproducibility per Part 12 Exception 2 (Retsch, Merck, MKN, Meiko, Tarrington House, EXTECH).
• Audit subagent (2026-06-02) flagged four Check-1 fidelity issues, all verified against the PDF: (1) “approximately 60,000 single food items” — verified against PDF page 1, which gives the methodology summary as “approx. 56,750 single food items”; the 60,000 figure is from section 4.2 (Purchasing) and represents gross purchasing volume; corrected to 56,750 with parenthetical note. (2) “Eight occurrence-data papers” — verified against Table 39 (page 64) which lists exactly seven entries in the occurrence-data block; corrected to seven. (3) “seven BfR opinions/communications” — verified against Table 39 (page 65) which lists exactly six entries; corrected to six. (4) Pistachio aflatoxin B1 framed as an “exceedance” — verified against Table 38 (page 63) which shows pistachios at 7.53 µg/kg against an 8 µg/kg ML (below the limit), parallel to the spiny-dogfish ndl-PCB notification-relevant near-miss; reframed to separate the actual buckwheat and chia-seeds exceedances from the pistachio near-miss. All four findings applied; zero false positives.

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.