da Silva et al. 2022 — Total mercury in Spanish baby food, fast food, and daily meals (preprint)

Summary

Spanish preprint study determining total mercury (THg) by direct mercury analyser (DMA-80) in twenty-eight composite meal samples representing three intake categories consumed in Valencia, Spain: thirteen commercial jarred baby foods for infants 6–12 months old, six children’s fast food menus, and nine adult canteen menus. Concentrations are reported per kilogram of dry weight and converted to daily intake per person using the measured moisture content (57–87 % w/w) and the fresh mass of each menu. The paper compares the resulting daily intakes to the Joint FAO/WHO Expert Committee on Food Additives (JECFA) Provisional Tolerable Weekly Intake (PTWI) for total mercury of 4 µg kg⁻¹ body weight (Tolerable Daily Intake of 0.57 µg kg⁻¹ bw day⁻¹). The dataset’s dominant signal is that mercury content tracks fish content of the meal: the highest THg concentrations are in fish-containing baby foods (39.8–41.9 µg kg⁻¹ d.w.), the highest fast-food value is in a single menu (67.7 µg kg⁻¹ d.w., the carrier of which the authors cannot attribute to any one component), and the single canteen menu containing grilled tuna reaches 638 µg kg⁻¹ d.w. — equal to 395.7 % of the adult TDI in a single meal. The paper concludes that baby food and most canteen menus pose low total-mercury risk, that children’s fast food can exceed 50 % of the TDI in a single meal at the lower body-weight assumption, and that the tuna meal alone contains half of the PTWI for adults.

Key numbers

Analytical performance (Results §3.2; Table 2):

• LOD 0.1 µg kg⁻¹; LOQ 0.3 µg kg⁻¹ (both derived from ten independent blanks using a 50 mg sample mass of Hg-free wheat flour).
• Relative standard deviation across triplicates spanning 0.43–638 µg kg⁻¹: 0.8–11 % (generally < 5 %).
• Recovery of spikes at 2.5 / 10 / 25 / 500 µg kg⁻¹: 98.6 ± 0.9 % / 99.2 ± 2.0 % / 102 ± 3 % / 104.9 ± 1.9 %.
• Certified reference material agreement (found vs certified, µg kg⁻¹): Chicken NCSZC73016 2.94 ± 0.18 vs 3.6 ± 1.5 (UΔ 1.5); Fish Protein NRC DORM-3 331 ± 6 vs 382 ± 60 (UΔ 60); Lobster Hepatopancreas NRC TORT-2 280 ± 58 vs 270 ± 60 (UΔ 90). All within combined uncertainty.

Total mercury by sample, dry-weight basis (Table 4; mean ± SD, n = 3 per sample):

Baby foods (range 0.57–41.9 µg kg⁻¹ d.w.; mean 15.5):

Children’s fast food menus (range 0.54–67.7 µg kg⁻¹ d.w.; mean 15.6):

Adult canteen daily menus (range 0.43–19.4 µg kg⁻¹ d.w., except DM8 at 638; mean 76.9 with DM8, 6.65 without DM8):

Effect of sample mass on direct Hg determination (Table 3; one trio per intake category):

• Baby Food 1 (Hg ≈ 21 µg kg⁻¹ d.w.): 20.98 ± 0.19 (50 mg single) → 19.76 ± 0.16 (50 + 50) → 19.5 ± 0.4 (100) → 19.86 ± 0.05 (50 + 50 + 50) → 19.45 ± 0.17 (150 mg). Differences ≤ 5 %; sample mass selected at 50 mg for the rest of the study.
• Fast Food 2 (Hg ≈ 0.5 µg kg⁻¹ d.w.): values drift from 0.56 ± 0.01 (50 mg) down to 0.40 ± 0.07 (150 mg) as the sample mass increases; the paper notes that masses ≥ 100 mg give worse results when sample THg is near the LOQ.
• Daily Meal 1 (Hg ≈ 14 µg kg⁻¹ d.w.): 14.7 ± 0.4 (50 mg) → 13.77 ± 0.14 (150 mg); ≤ 7 % drift across the mass range.

Reference intake parameters (Methods §2.5; WHO 2011):

• JECFA PTWI for total Hg: 4 µg kg⁻¹ body weight per week, equivalent to a TDI of 0.57 µg kg⁻¹ bw day⁻¹.
• Age-group TDIs in µg day⁻¹ used by the paper: infants 7–12 months 5; toddlers 13–24 months 7.4; children 3–7 years 11; children 7–12 years 20; adults 34.
• Body weights: 9 kg (7–12 mo), 13 kg (13–24 mo), 19 kg (3–7 yr), 35 kg (7–12 yr), 60 kg (adults).
• Daily Meal 8 contributes 134 µg Hg in a single meal versus the PTWI of 240 µg per 60-kg adult per week — i.e., a single tuna-containing meal carries 56 % of an adult’s weekly tolerable intake.

Cross-country baby food comparison (Table 5; literature pull):

• Spain (this work, n = 13): 0.57–41.9 µg kg⁻¹; mean 15.5.
• China, Spain, UK, USA (Carbonell-Barrachina et al. 2012): China < 4–15 (mean < 4), Spain < 4–21 (mean 5), UK < 4–10 (mean 5), USA < 4–6 (mean < 4).
• Italy / Slovakia / Spain / Sweden (Pandelova et al. 2012): 2.7–29.9 µg kg⁻¹; mean 9.2.
• Portugal (Martins et al. 2013): < 0.10–19.6 µg kg⁻¹; mean 0.40.
• Czech Republic / Hungary / Slovakia (Tóth et al. 2014): 0.3–10.2 µg kg⁻¹; mean 1.82.
• Italy (de Roma et al. 2017): mean 1.69 µg kg⁻¹ (range not given).

Evidence Fitness

This source supports occurrence evidence for total mercury in Spanish jarred baby foods (n = 13, sampled from local markets in Valencia), and exposure context for children’s fast food menus and adult canteen menus (smaller n, single-city draw). The dataset is reconstructable down to the individual sample because Table 4 reports each composite meal’s concentration, daily intake per person, and % TDI separately rather than only the group mean.

Limitations the source itself acknowledges:

• Total mercury only; no methylmercury speciation, so the high tuna value in Daily Meal 8 cannot be partitioned into the MeHg fraction directly responsible for the EFSA developmental TWI.
• Each “menu” is a single composite sample (one purchase × three analytical replicates), not a triplicate sampling of independent purchases of the same menu type. Sample-to-sample variability of the same menu type cannot be estimated from this dataset.
• TDI percentages are computed against the single meal in question, not the rest of the day’s diet; the paper explicitly states “the studied menus did not represent the food intake for a whole day.”
• Fast-food samples (n = 6) share an almost identical composition (burger + chips + ketchup + yogurt or milkshake + drink), and the paper notes that variation in THg “could not be assigned to any of the components, and no relation with the brand was found” — i.e., the source itself flags an unexplained 100-fold spread (0.54–67.7 µg kg⁻¹ d.w.) within a nominally homogeneous menu.
• Status as a preprint: posted on Research Square 2022-05-05, DOI 10.21203/rs.3.rs-1614710/v1; the Research Square header explicitly states “Preprints are preliminary reports that have not undergone peer review.” No subsequent peer-reviewed version has been located by the wiki ingest at the access date below.

The source can be used as occurrence evidence for the broader baby-food product page and as one input to the fish-containing baby food category. It cannot, on its own, set a percentile for either category at HMTc-grade pooling levels because the sample size is small and single-purchase. For the high-Hg tuna meal it is illustrative rather than definitive, because the dish-level value (638 µg kg⁻¹ d.w.) reflects whatever tuna lot was on the canteen menu that day.

Methods

Sampling. Twenty-eight composite meal samples purchased in Valencia, Spain. Thirteen jarred baby food samples from different brands available in Spain were obtained at local markets and classified by meat / fish / vegetable content. Six children’s fast food menus were purchased from different commercial brands in Valencia town and shopping centres; each menu was composed of a hamburger bun, beef hamburger, cheddar cheese, gherkin, ketchup, mustard, a regular portion of French fries, yogurt or milkshake, and a soft drink. Nine adult daily canteen menus were purchased at the University of Valencia canteens; each menu was composed of a starter, main dish, dessert, and 50–70 g bread. Twenty-eight samples total; Table 1 lists each menu’s composition and fresh mass. Moisture ranged from 57 to 87 % w/w.

Sample preparation. Samples were crushed, homogenised, frozen at −20 °C and freeze-dried for ≥ 48 h at 0.05 mBar to remove water and pre-concentrate Hg, then homogenised again in a domestic mixer (Braun, Kronberg, Germany) and stored in polyethylene tubes. No acid digestion was performed (the DMA-80 accepts solid samples directly).

Instrumentation. Direct Mercury Analyzer DMA-80 (Milestone, Sorisole, Italy). Solid sample (50 mg, weighed into a nickel crucible) was dried and thermally decomposed in the instrument; the decomposition products were carried by an oxygen flow over a catalyst that completed sample oxidation and trapped halogens and N/S oxides; the remaining gas passed through a gold amalgamator that collected Hg⁰. The amalgamator was heated to 850 °C, the released Hg⁰ was transported into the absorption cell, and total mercury was quantified by atomic absorption at 253.7 nm. Two analytical curves were used to span low (0–20 ng) and high (20–1000 ng) Hg masses (different optical path lengths; r² > 0.99 in both ranges). Oxygen was used as the carrier gas. Operating conditions are listed in supplementary Table S1.

Calibration. Aqueous standards prepared from Hg(II) 1000 µg mL⁻¹ Merck standard solution and ultrapure water (Milli-Q, 18.2 MΩ cm). Calibration verified each session with certified reference material.

Sample-mass study (Table 3, Results §3.3). Three representative samples (Baby Food 1, Fast Food 2, Daily Meal 1) were analysed at 50, 100, and 150 mg, and also as 2 × 50 mg and 3 × 50 mg amalgamator stacks, to evaluate matrix loading effects on accuracy. A 50 mg single load was selected for the full set; the authors note that masses ≥ 100 mg degrade accuracy at near-LOQ concentrations.

Reference materials used for QC. Chicken NCSZC73016 (China National Analysis Center for Iron and Steel, Beijing); Fish Protein NRC DORM-3 and Lobster Hepatopancreas NRC TORT-2 (National Research Council of Canada, Ottawa). Prior validation by the same research group also covered Fucus IAEA 140-TM, Coal Fly Ash NIST 1633b, and NIES Rice 10-a/b/c (cited as da Silva et al. 2014).

Risk-assessment calculation. Daily intake (DI, µg day⁻¹) = DCM × MMC × 1000, where DCM is daily consumption of the menu (g day⁻¹) and MMC is the mean Hg concentration (µg kg⁻¹). %TDI = DI / (4 µg kg⁻¹ × body weight ÷ 7 days) × 100 with body weights from de Lara et al. 2010 and Hernández-Martínez et al. 2013. Both DI and %TDI are reported per-single-meal, not per-day.

Speciation. Total mercury only. No MeHg / iHg / Hg⁰ separation. The paper explicitly notes that “mercury dietary intake mainly comes from fish and shellfish consumption” and treats fish-containing menus as expected high-THg contributors, but does not measure the methylated fraction.

Implications

• Occurrence — Spanish jarred baby foods. Contributes 13 sample-level THg values on a d.w. basis spanning 0.57–41.9 µg kg⁻¹, with explicit composition descriptors (fish / meat / vegetable) that allow downstream routing to fish-containing-baby-food vs vegetable-only baby food sub-rows. Mean 15.5 µg kg⁻¹ d.w. sits at the upper end of the European range reported in Table 5 (Portugal 0.40; Czech / Hungary / Slovakia 1.82; Italy 1.69; pooled European range across Pandelova 2012 et al. 2.7–29.9, mean 9.2). Seven of the thirteen baby food jars contain a fish component (BF1, BF3, BF5, BF6, BF11, BF12, BF13; THg 11.5–41.9 µg kg⁻¹ d.w., mean 23.93) and sit well above the two pure-vegetable jars (BF7, BF8; 0.57 and 0.63 µg kg⁻¹ d.w., mean 0.60). The remaining four jars are meat-or-poultry-bearing without fish (BF2 lamb 21.8; BF4 beef 2.51; BF9 chicken 2.85; BF10 chicken-broth 7.38; range 2.51–21.8, mean 8.64) and sit between the two extremes.
• Occurrence — children’s fast food menus, Valencia. Six near-identical composite menus span a 100-fold range (0.54–67.7 µg kg⁻¹ d.w.) with no source attribution the authors could resolve. This is a small but relevant occurrence input for the broader “prepared meal / fast food” category, with the caveat that all six samples were drawn from a single city and a near-uniform menu composition.
• Occurrence — adult canteen menus, single canteen system. Nine composite menus span 0.43–19.4 µg kg⁻¹ d.w. plus one outlier at 638 µg kg⁻¹ d.w. carrying grilled tuna; the outlier is a single dish-level value rather than a population estimate. Useful as illustrative high-Hg event evidence; not adequate for percentile-setting on its own.
• Tuna-containing dish as PTWI-fraction signal. Daily Meal 8 contributes 134 µg Hg in one meal — i.e., 56 % of the JECFA PTWI for a 60-kg adult (240 µg/week). This is consistent with the broader literature pull the paper cites (tuna 222–776 µg kg⁻¹ in Martorell et al. 2011; swordfish 856 µg kg⁻¹ in González et al. 2019) and is illustrative of the dominant role predatory marine fish play in adult Hg exposure.
• Method. The DMA-80 direct mercury analyser is reproducible and accurate against three independent certified reference materials at this lab’s concentration range; LOD 0.1 µg kg⁻¹ and LOQ 0.3 µg kg⁻¹ on 50 mg dry mass loads are usable for category-level occurrence work on foods at the µg kg⁻¹ d.w. scale.

Provenance notes

• Source is a Research Square preprint, DOI 10.21203/rs.3.rs-1614710/v1, posted 2022-05-05 under CC BY 4.0. Preprint status is explicit on the manuscript header; no subsequent peer-reviewed version located at ingest date. Evidence tier B reflects preprint status; should an indexed peer-reviewed version appear, this page should be superseded.
• PDF retrieved via the discover skill into raw/Manual Fetch Discovery/silva2022-total-mercury-baby-food.pdf. SHA-256 8fb127b5a0871d10a7f53f1d17dce4fc3b1f2e75b86a204ee3adab0338d4a49c. Twenty-two pages including five tables and a reference list. Tables 1 and 4 are rendered as proper tabular elements in the PDF; values were read directly from the rendered cells.
• No supplementary files were retrieved; the manuscript references SupplementaryMaterial.docx (DMA-80 operating conditions, Table S1) and GraphicalAbstract.docx as available downloads on the Research Square landing page, neither pulled into this ingest.

Wiki pages this source may touch

• mercury-total
• baby-food
• fish
• baby-food
• fish-containing-baby-foods
• efsa-mercury-twi

Verification notes

• Speciation discipline (CLAUDE.md Part 14). Paper measures total mercury only (DMA-80 thermal decomposition + gold amalgamation + AAS at 253.7 nm). Frontmatter metals: therefore uses tHg, not MeHg; the metal page slug referenced in the body is metals/mercury-total. The dominant fish-driven signal in baby foods and the tuna-driven Daily Meal 8 result would, in a speciated study, be presumed largely methylmercury, but this paper does not measure the methylated fraction and the wiki page should not impute it.
• Brand firewall (CLAUDE.md Part 12, strict reading). The source itself does not name any baby food, fast food, or canteen brand by name; samples are referenced only as “Baby Food 1”, “Fast Food 1”, “Daily Meal 1” with composition descriptors (sole with white sauce, hake and white sauce, grilled tuna with steamed potatoes, etc.). No brand-by-brand attribution exists in the source to strip; product-form descriptors are preserved as the source uses them. Method-section instrument and reference-material vendor names (Milestone DMA-80, Merck Hg standard, Milli-Q water, NRC DORM-3, NRC TORT-2, NCSZC73016) are scientific reproducibility detail per the locked 2026-05-17 exception, not brand-firewall hits.
• Wiki/HMTc firewall (CLAUDE.md Part 2). No HMTc threshold proposals appear on this page; the ## Implications section reports what the dataset contributes to occurrence work and explicitly does not propose certification values. No consumer-audience translation appears; the TDI / PTWI fractions are reported as the source reports them (paper-internal exposure assessment), not as risk advisories.
• Basis discipline. All sample concentrations in Table 4 and on this page are reported per kilogram of dry weight (freeze-dried, as the analytical sample). The source’s daily-intake column converts to fresh-weight-equivalent µg day⁻¹ via the per-sample fresh mass and moisture content in Table 1. Both columns are preserved here without unit reconversion; downstream pooling that needs as-consumed values must apply the per-sample moisture fraction from Table 1 rather than assume the dry-weight value.
• Routing scope. The thirteen baby foods are routed to the umbrella products/baby-food slug rather than to soy / non-soy / cereal-based / puree subcategories because the source classifies by ingredient content (meat / fish / vegetable) rather than by jarred-baby-food taxonomy. Five of thirteen contain fish (BF1, BF3, BF5, BF6, BF11, BF12, BF13 — actually seven; see correction below) and are additionally routed to products/fish-containing-baby-foods. No fast-food or canteen-meal product slug exists in the current taxonomy and these intake categories are flagged in the verification notes only, not routed.
• Fish-containing-baby-food subset (revised after subagent audit). Counting fish-containing baby foods from Table 1: BF1 (sole), BF3 (hake with rice), BF5 (monkfish with vegetables), BF6 (whiting with vegetables), BF11 (sea bass with vegetables), BF12 (hake with peas and rice), BF13 (hake with white sauce) — that is seven of thirteen, with THg values 20.98, 15.89, 19.22, 11.5, 39.8, 18.23, 41.9 µg kg⁻¹ d.w. (range 11.5–41.9; mean 23.93). The pure-vegetable subset is BF7 (mixed vegetables) and BF8 (cream of green beans with potatoes) — two of thirteen, values 0.63 and 0.57, mean 0.60. The remaining four jars are meat-or-poultry-bearing without fish: BF2 (lamb 21.8), BF4 (beef 2.51), BF9 (chicken 2.85), BF10 (carrots-and-rice in chicken broth 7.38) — range 2.51–21.8, mean 8.64. The Implications section now reflects this three-way breakdown directly.
• No ingredient or product page creation. Per CLAUDE.md Part 10, this ingest does not propose new ingredient or product pages. Two slugs the dataset would otherwise have touched are not in the current taxonomy: a tuna ingredient slug (only canned-tuna and swordfish exist; the Daily Meal 8 result is fresh grilled tuna, not canned) and any “prepared-meal” / “fast-food” / “canteen-meal” product slug for the fast food and daily canteen intake categories. The high tuna value is documented in the body but not forced into the routing layer.
• Fresh-context audit applied 2026-06-09. A read-only subagent audit (Claude Code, fresh context) returned REVISE. Findings verified against the source PDF and applied:
  • ❌ Adult-meal mean excluding DM8 was miscomputed. The Key numbers caption read “otherwise 7.5”; recomputing from Table 4 gives (14.7 + 19.4 + 3.14 + 0.95 + 0.48 + 0.43 + 0.63 + 13.5) / 8 = 53.23 / 8 = 6.65 µg kg⁻¹ d.w. Corrected to 6.65.
  • ❌ Implications section “5 of 13” and “4 of 13” subset counts were stale. The prior version of the Verification notes asserted the Implications had been corrected to the 7 / 2 / 4 jar breakdown, but the Implications text itself still carried the old 5 / 4 split with mean ≈ 27 µg kg⁻¹ d.w. for the fish subset and mean ≈ 2.4 for the vegetable subset. Now rewritten in the Implications bullet to match the Table 1 / Table 4 jar-by-jar breakdown: 7 fish-bearing (mean 23.93), 2 pure-vegetable (mean 0.60), 4 meat-or-poultry-without-fish (mean 8.64).
  • ❌→⚠️ JECFA slug removed. The original ## Wiki pages this source may touch block listed regulations/jecfa-methylmercury-ptwi. The source actually invokes the JECFA Provisional Tolerable Weekly Intake of 4 µg kg⁻¹ bw week⁻¹ for total mercury (Methods §2.5, citing WHO 2011 / JECFA 72nd report), not the 1.6 µg kg⁻¹ bw week⁻¹ MeHg PTWI. No JECFA total/inorganic-Hg PTWI slug exists in the current taxonomy snapshot (only jecfa-methylmercury-ptwi for mercury under JECFA); the incorrect slug was therefore dropped rather than replaced. This is a slug-vocabulary gap that should be opened as a regulations/jecfa-mercury-ptwi proposal in a future taxonomy review.
  • ⚠️ “Macaroon” silently normalized to “Macaroni” in DM5 composition (Key numbers Daily Meal 5 row). PDF Table 1 prints “Macaroon with tomato” (p. 18), which appears to be an English-edit typo in the manuscript (“macaroni con tomate” is the standard Spanish canteen dish). The wiki page renders it as “Macaroni” without flagging the source typo. Flagged here for transparency; not reverted, since “Macaroni” is the dish the source’s Spanish-canteen context plainly intends.
  • ⚠️ DM2 and DM4 salad-ingredient lists truncated. Table 1 lists the DM2 salad as “lettuce, carrots, ham, soy, cheese, and mayonnaise” and the DM4 salad as “lettuce, tomato, corn, carrots, eggs, cucumber, pepper, soy, and olives”; the Key numbers table on this page elides both as “Mixed salad” for column-width reasons. The truncation does not affect the mercury values reported for those rows, but the presence of soy and ham/eggs in these adult-canteen salads is dropped from the wiki view of the composition column.
  • Subagent verdict: REVISE. All three ❌ findings verified correct against the PDF and applied; both ⚠️ findings verified and documented here as transparency notes (DM5 normalization left as-is; DM2/DM4 truncation left as-is). No false positives in this audit cycle.

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.