EL Daouk et al. 2020 — Aluminum exposure from food in the Lebanese population

This is the first national-scope study of dietary aluminum (Al) exposure in Lebanon. The authors combined a targeted retail-market survey of 97 food items measured by flame atomic absorption spectrometry (FAAS) with a self-administered Electronic Food Frequency Questionnaire (E-FFQ) completed by 167 Lebanese adults to estimate Daily Dietary Exposure (DDE) and Provisional Tolerable Weekly Intake (PTWI) for Al. Food selection followed the second French Total Diet Study (EAT2) and Lebanese Food Based Dietary Guidelines, with explicit inclusion of traditional Lebanese foods cooked in aluminum utensils (kanafeh, maamoul, foul, falafel, shawarma) and acidic foods stored in metal cans or aluminum foil. Mean Al concentration across the 97 items was 3.56 ± 2.08 mg/kg fresh weight (range 0.14–9.37 mg/kg), with the highest group-mean concentrations in vegetables, sauces and condiments, candies, ready meals, potatoes, and coffee/cocoa/tea. Mean DDE was estimated at 4341.18 µg/day, giving a 50th-percentile Dietary Weekly Intake (DWI) of 0.50 mg/kg bw/week at an assumed 60 kg body weight — half of the JECFA/EFSA PTWI of 1 mg/kg bw/week — and a 95th-percentile DWI of 1.01 mg/kg bw/week, at the threshold. Top contributors to ingested Al by total daily food mass were soft drinks, fresh tomatoes, tea, ice cream, Turkish coffee, and fresh lettuce. The authors attribute the bulk of Al loading to packaging (cans, foil, capsules), cookware migration into acidic foods, and food additives, with secondary contribution from soil/irrigation-water uptake into vegetables.

Key numbers

All concentrations in mg/kg (= ppm) fresh weight unless noted. Exposure values in µg/day or mg/kg bw/week with a fixed 60 kg default body weight (paper’s assumption, applied to both sexes; no Lebanese-specific bw available). Triplicate analysis per item. Wet/dry basis: the authors digested 5 g fresh weight of solid matrices, so concentrations are on a wet (fresh-weight) basis.

Overall corpus

Per-food-group mean Al (mg/kg fresh weight), top contributors and selected lower-ranked groups

The paper reports these as the maximum (highest-Al) item within each group rather than a group-average across heterogeneous items; the abstract and conclusion describe them as the group’s representative high value. Ordering follows Figure 1.

Highest individual ingestion rates (g/day) — exposure-driving foods

These are E-FFQ-derived consumption rates multiplied by portion size; they identify the foods doing the most exposure work regardless of per-item Al concentration.

Contact-pathway classification of the 97 items

Population exposure (DDE / DWI)

E-FFQ participant DWI by subgroup (mg/kg bw/week)

None of the gender, age-group, or region differences reached p < 0.05; reported here for completeness.

Cross-country comparison reported by the authors (DDI, µg/kg bw/day)

Methods (brief)

Survey instrument. Customized self-reported Electronic Food Frequency Questionnaire (E-FFQ) hosted on the Curve® web platform, listing 67 food items with portion sizes referenced to the Lebanese Food Based Dietary Guidelines. Frequency conversion followed the EPIC factor scheme (0 to once a day across five bins). Disseminated by messenger-service link segmented by age and proportionally to the five Lebanese provinces (Beirut, Mount Lebanon, South including Nabatiyeh, North including Akkar, Bekaa including Hermel). N = 167 respondents, ages 18–64. Data analysis in SPSS 25 with ANOVA, one-sample and independent t-tests; p < 0.05 considered significant. Missing values < 10%.

Sampling. 105 packed (n = 70) and unpacked (n = 35) food items randomly collected May–September 2018 from Beirut market; 97 retained for analysis (8 discarded for turbidity). Items selected from the 2nd French Total Diet Study (EAT2) results filtered for availability in the Lebanese market and inclusion of traditional Lebanese dishes prepared in aluminum utensils. Plastic bags and plastic cutlery used throughout sampling and handling to avoid metal cross-contamination. Solid matrices: 5 g fresh weight, longitudinally mid-cut, homogenized, placed in porcelain crucibles. Liquid matrices: directly weighed in porcelain crucibles.

Digestion. Dry-ash method (WiseTherm muffle furnace, Merck/Daihan): solids ramped to 550 °C for 3–4 h; liquids ramped through 120 °C / 200 °C / 350 °C / 500 °C / 550 °C at 0’ / 60’ / 120’ / 180’ / 210’. Ash transferred to 50 mL PTFE beakers with 1:1 v/v H₂O₂ (30%) : HNO₃ (65%) added in 15 mL portions; heated on hotplate until clear (15–30 min); cooled, sonicated, 0.45 µm filtered into 20 mL graduated tubes; stored at 4 °C in darkroom. AOAC standard protocols followed. Analysis performed at the Industrial Research Institute (IRI) laboratory, accredited by ANAB-ANSI.

Instrumentation. Flame Atomic Absorption Spectrometry (FAAS) on Shimadzu AA-6800 with ACS 6100 autosampler. Aluminum lamp wavelength 309.3 nm, slit width 0.7 nm, nitrous oxide (N₂O) oxidant gas. Eight-step temperature/ramp/heat program tabulated in source (60 °C → 120 °C → 250 °C → 900 °C × 3 → 2600 °C × 2 with internal N₂ flow 0.10–1.00 L/min). Ultra-pure water from Evoqua LaboStar 2 (Siemens, UK); HNO₃ and H₂O₂ from VWR (BDH, France); glassware soaked overnight in 20% v/v HNO₃ then rinsed.

Calibration and QC. External calibration curve 0.1–10 ppm from a multi-element 500 µg/mL standard in 2% HNO₃ + trace HF (High Purity Standards Greyhound, USA). Linearity r² > 0.9942. Recovery determined by spiking at 2, 5, 10, 15, 20 ppm: 87.52–88.1%. LOD = 0.04 ppm and LOQ = 0.12 ppm (from 12 blank measurements; LOD = 3 × SD, LOQ = 10 × LOD). Accuracy > 97% against a Certified Reference Material at 0.4 ppm Al. Precision (CRM reproducibility SD) 95.5%. Expanded measurement uncertainty 2.00 ± 0.05 ppm. Control samples run alongside test samples.

Exposure calculation. DDE (mg/day) = Σ [Al content (mg/kg) × food intake (g/day)] / 1000. PTWI estimated by multiplying DDE × 7 and dividing by 60 kg default body weight (no Lebanese-specific bw available; chosen for comparability with French EAT2). 50th-percentile (typical) and 95th-percentile (upper-end) consumption rates used.

Limitations

• Sample size per item is small: many food items represented by n = 1 individual product purchase (the parenthetical count in Table 3 is items, not replicates; the triplicate refers to repeated FAAS measurements of the same item). The 5 mg/kg group-cutoff descriptive statistics are correspondingly underpowered for generalisation to the national supply.
• All 105 items purchased from Beirut market; the four other Lebanese provinces (Mount Lebanon, South, North, Bekaa) are not represented in the food-sampling arm, although their populations are represented in the E-FFQ exposure arm.
• 60 kg default body weight is the FAO 1962 Lebanon nutrition-survey figure assumed for both sexes; no contemporary Lebanese-specific body weight distribution was available. DWI estimates scale inversely with bw.
• E-FFQ is subject to recall and social-desirability bias; the authors note that under-reporting and over-reporting cannot be excluded.
• Food selection is not exhaustive — it is the EAT2-filtered subset of items expected to carry Al, not a full national diet. The DDE total is therefore a lower bound on whole-diet Al intake; foods omitted because they were not in the EAT2 high-Al list contribute zero in this accounting.
• Analytical method is single-element FAAS rather than ICP-MS; FAAS is standard for Al but does not provide species information (the paper measures total Al).
• Several traditional Lebanese dishes (Kanafeh, Maamoul) were prepared in aluminum trays as part of normal commercial preparation; the contribution of cookware migration during preparation cannot be separated from contributions due to additives or packaging.
• The cyclic figure (Figure 1) reports per-group highs that are sometimes interpretable as group means and sometimes as the highest item within the group; the abstract and Table 3 numerical reconciliation distinguish these where possible.
• 8 of 105 items (7.6%) were discarded for analytical turbidity; the authors do not characterise these items or whether their exclusion biases the food-group coverage.
• Single-year sampling (May–September 2018) does not capture seasonal or interannual variation in vegetable Al loading from soil and irrigation water.

Implications

For HMI evidence corpus. This study contributes Lebanese-population occurrence data for Al across a structured set of 18 retail food groups, including categories underrepresented in the existing corpus: Arabic sweets, charcuterie (Levantine processed-meat formulations), ready meals (shawarma, falafel, basterma, makanek), Turkish coffee, packed and powdered fruit juices, and traditional Al-utensil-prepared dishes. It also adds national-scale exposure modelling (mean DDE 4341.18 µg/day; 50th-percentile DWI 0.50 mg/kg bw/week) referenced against the JECFA/EFSA Al PTWI of 1 mg/kg bw/week — the lower-tier of the literature distribution for Mediterranean / Eastern Mediterranean Al exposure, lower than Spain, Canada, France’s high-end estimates and lower than Italy/UK at the central tendency but at the same order of magnitude.

For Al-pathway coverage. Three loading routes are represented in the dataset and worth carrying through to downstream pages: (i) acidic-food contact with metal cans and aluminum foil (soft drinks, canned juices, canned vegetables, condiments, canned fruit, sardines), (ii) cookware migration during preparation in aluminum trays and pots (Arabic sweets, Turkish coffee, fried foods, barbecue), (iii) food additives functioning as anticaking, firming, raising, and stabilising agents (candies, processed cheese, packed melted cheese, baking-derived items). The 9-cookware-items / 63-packaging-items / 5-foil-items / 20-unspecified split is one of the cleaner pathway tabulations in the literature.

For methodology comparison. FAAS with 309.3 nm Al lamp, N₂O oxidant, and dry-ash digestion at 550 °C is the standard non-ICP Al protocol; LOD 0.04 ppm and LOQ 0.12 ppm are within typical FAAS-Al ranges. Recovery at 87.5–88.1% is on the lower end of acceptable per AOAC; downstream synthesis should attach a method-bias caveat when this study is one of the top contributors to a per-ingredient Al value. The Certified Reference Material is reported at 0.4 ppm Al with > 97% accuracy but the CRM identity is not named, weakening third-party reproducibility.

Cross-source compatibility. The Lebanese DDI of 71.4 µg/kg bw/day sits centrally among the cross-country range the authors themselves cite (Australia 36, France 40.3, Italy 60, UK 71, Lebanon 71.4, Canada 105–173, Spain 170). The paper reports the Lebanese estimate as “similar to findings from Italy and the UK (60 vs 71 µg/kg bw/day)“.

Wiki pages this source may touch

• aluminum
• lettuce
• tomato
• carrots
• tea
• coffee
• cocoa
• chocolate
• bread
• white-bread
• whole-milk
• yogurt
• butter
• potatoes
• potato-chips
• ketchup
• cucumbers
• lentils
• canned-corn
• cinnamon
• soft-drinks-carbonated-beverages
• tea
• condiments-general
• canned-vegetables
• canned-fruit
• bread-and-baked-goods
• processed-meats
• cookware-metal-alloy
• food-packaging-foils-wraps
• food-packaging-cans-lids
• efsa-aluminium-twi
• atsdr-aluminum-mrls

Verification notes

• 2026-06-08 — Fresh ingest from Manual Fetch Discovery. The discovery-time filename (el2020-matcha-heavy-metal.pdf) is misleading: the paper is not about matcha. It is the El Daouk et al. 2020 Lebanese national Al-exposure study published open access (CC BY 4.0) in Toxicology Reports 7:1025–1031 (doi 10.1016/j.toxrep.2020.08.018). The misleading discovery handle is preserved as raw_handle: MFD_el2020-matcha-heavy-metal to keep the filesystem trail intact (precedent: afzal2024-livestock-heavy-metals-review which retained MFD_afzal2024-camel-milk-heavy-metals despite not being about camel milk); the wiki cite-key reflects the paper’s actual content (el-daouk2020-aluminum-food-lebanon).
• 2026-06-08 — First author surname is the compound “EL Daouk” (as printed in the paper byline and at the corresponding-author email sdaouk@ul.edu.lb); the cite-key uses the hyphenated form el-daouk to keep the surname-year prefix unambiguous.
• 2026-06-08 — Frontmatter metals: [Al] only. Although the paper discusses other heavy metals in its background literature review (citing Lebanese studies on Pb, Cd, and trace elements by Nasreddine 2010, Lebbos 2019, Al-Chaarani 2009, Fadel 2016, Najoie 2019), the present study measures only aluminum. The other elements are not in scope.
• 2026-06-08 — Brand firewall: the paper does not name food brands for the 97 items (categories like “ketchup”, “packed melted cheese”, “in sachet instant coffee” are food-form descriptors, not brand names). The platform name “Curve®”, instrument vendors (Shimadzu, Merck/Daihan, Evoqua/Siemens), reagent vendors (VWR/BDH, High Purity Standards Greyhound), and the analytical lab (Industrial Research Institute, ANAB-ANSI-accredited) are retained per Part 12 Exception 2 (scientific-method vendor/material names).
• 2026-06-08 — Food-group concentrations in Figure 1 are reported as group-high values in the abstract and conclusion; Table 3 lists individual food items within each group with item-specific ingestion rates rather than per-item Al concentrations. The Key numbers tables above preserve this distinction (group-high column for the Figure 1 ranking; ingestion-rate column for Table 3).
• 2026-06-08 — Total of 167 E-FFQ respondents reconciles across the gender breakdown (62 + 105 = 167) but the age and region breakdowns sum to 166 (age: 92 + 30 + 27 + 17 = 166; region: 74 + 24 + 22 + 29 + 18 = 167). The Beirut count of 74 in the region table includes the missing one age-group response. Reproduced as-printed without adjustment.
• 2026-06-08 — The paper’s stated PTWI value of “0.50 mg/kg body weight (60 Kg/person)” in the abstract is the 50th-percentile DWI in mg/kg bw/week (paper text and Table 2 confirm “mg/kg bw/day” label is a misprint for “mg/kg bw/week”; the DDE-to-PTWI conversion in Section 2.3.4 multiplies by 7 days and divides by body weight). The 1 mg/kg bw/week PTWI reference is the post-2008 JECFA value that replaced the withdrawn 7 mg/kg bw/week PTWI, with a 2 mg/kg bw additive-source carve-out added in 2011.
• 2026-06-08 — Journal-name display in the article header reads “The End-to-End Journal” with link journal homepage: www.elsevier.com/locate/ENDEND, which appears to be an ScienceDirect template artifact; the paper itself is volume 7, pages 1025–1031 of Toxicology Reports (ISSN 2214-7500), confirmed by the running header, DOI prefix (10.1016/j.toxrep), and the journal’s own table of contents. Frontmatter uses “Toxicology Reports”.
• 2026-06-08 — Matrices vocabulary proposals: this is a TDS-style multi-matrix study, and the matrices slugs used (mixed-diet, vegetables, soft-drinks, sauces-condiments, candies, ready-meals, potatoes, coffee-cocoa-tea, dairy-products, processed-cheese, charcuterie) are study-derived broad food-group slugs that may not all be in the closed matrices vocabulary in docs/gpt-collaboration/system-prompt.md. They mirror the paper’s 18 food-group stratification (Section 2.3.5) and Figure 1 category names. Flagged here as proposals for the matrices vocabulary if not already present; the routing audit treats them as bare strings (not wikilinks), so they do not block routing.
• 2026-06-08 — Audit subagent (general-purpose) flagged that the “Per-food-group mean Al” Key-numbers table omitted Charcuterie. Verified against PDF p.1027 §3: “between 5 and 6 (ready meals, candies, charcuterie); 4 and 5 (desserts, cream, jam, cake, croissant, biscuits, sauces & condiments, vegetables, processed cheese and potatoes); 3–4 (chicken, fish, meat, soft drinks, legumes, coffee, cocoa and tea); 2–3 (juices, Arabic sweets, bread, pastry, …)“. Finding confirmed correct; added “Charcuterie | 5–6 band” row. Note: the paper’s 2–3 band passage redundantly relists chicken/fish/meat, soft drinks, legumes, coffee/cocoa/tea — a transcription duplication in the source; the table prefers the first (3–4 band) classification per the source’s own group-mean ranking.
• 2026-06-08 — Audit subagent flagged the Implications §“Cross-source compatibility” sentence “the Mediterranean clustering at 60–71 µg/kg bw/day across three independent national studies is a multi-source convergence worth flagging at the metals/aluminum synthesis pass” as wiki-side synthesis prescription rather than source-reporting (Part 2). Finding confirmed correct; rewrote to report only what the paper itself states (“similar to findings from Italy and the UK (60 vs 71 µg/kg bw/day)”) and dropped the synthesis-pass directive.
• 2026-06-08 — Paper-internal inconsistency in Methods Section 2.3.1 + Table 1: the text states “Flame Atomic Absorption Spectrometry (FAAS) … Nitrous oxide (N₂O) was used as oxidant gas”, but Table 1 (titled “FAAS operating parameters for the determination of Al, Shimadzu cookbook”) lists an eight-step temperature program ramping from 60 °C to 2600 °C with internal N₂ flow 0.10–1.00 L/min — a graphite-furnace AAS (GFAAS) atomizer program, not a flame AAS instrument profile (which does not have a programmable temperature ramp). The Shimadzu AA-6800 supports both flame and graphite-furnace modes, but the table parameters describe the GFAAS pathway while the text describes a flame N₂O-acetylene Al determination. The recovery, LOD, LOQ, and accuracy values reported in Section 2.3.3 are consistent with FAAS-Al performance; we reproduce both descriptions as printed and note the inconsistency here rather than silently picking one. Not a stop condition: the analytical results stand on their reported recovery/LOD/precision against CRM regardless of which atomization pathway was used.

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.