Lynch 2014 — Comprehensive meta-analysis of inorganic arsenic in food (seaweed is the highest-inAs group)

Lynch and colleagues compiled 134 studies into a >6500-data-point database to estimate mean speciated-arsenic concentrations across food groups, focused on inorganic arsenic (inAs). This is a secondary source (literature meta-analysis), not a primary measurement study; its values are aggregates of the underlying primary studies and so are recorded here as literature-wide context, not as poolable occurrence data (pooling them would double-count the primary sources). Its central finding for the Category 6 seaweed row: seaweed/algae had the highest mean total inorganic arsenic of any food group, 11,000 µg/kg (n=312, range <14–120,000), with seaweed-containing products far lower at 260 µg/kg (n=30). The review also documents extensive data-quality caveats (extraction efficiencies often <80% for algae, high non-detect fractions, combined dry/wet-weight reporting) that bear on how seaweed arsenic data should be interpreted.

Key numbers

Meta-analytic means (µg/kg unless noted); these are literature aggregates, NOT new measurements and NOT poolable.

Seaweed (the row-relevant group; Table 3.3):

• Total inorganic As: mean 11,000 µg/kg, n=312, range <14–120,000 — the highest mean inAs of any food group analysed.
• inAs³⁺: 1200 (range <0.1–6700), n=37; inAs⁵⁺: 21,000 (range <17–110,000), n=105.
• MMA: 2200 (n=58); DMA: 2200 (range <7–46,000), n=103; Total As: 39,000 (range 31–150,000), n=421.
• Seaweed products (foods with seaweed as an ingredient): total inAs mean 260 µg/kg, n=30, range <14–1700; % inAs of total ranged 1–91%.
• % inorganic As of total in seaweed/algae samples ranged 0–86%; 18 of 312 inAs values were non-detect (6%) (Table 3.5).

For cross-food context (Table 3.3, same meta-analysis):

• Rice: total inAs mean 130 µg/kg, n=1126; rice products 760 µg/kg, n=151.
• Seafood: total inAs 130 µg/kg, n=835 (total As 10,000 µg/kg, n=2417); Fish total inAs 45 µg/kg.
• Infant rice cakes/cereal: total inAs ~92 µg/kg (range 8–220), n=65 — highest among infant foods.

Additional source-reported facts:

• The review contrasts its seaweed mean (11,000 µg/kg inAs, ≈25% of total As) with JECFA (2011), which assumed inAs = 50% of total As in hijiki and <15% in other seaweed; the meta-analysis mean is “much lower than the assumption used in the EFSA analysis.”
• The maximum compiled total inAs across all foods was ~120,000 µg/kg, in dried seaweed/algae.
• Sensitivity analysis (Table 4.1): restricting to dry-weight-only seaweed values changed the seaweed inAs mean by −38% vs the combined dry/wet main analysis (wet-weight-only: no change; seaweed products dry-only: +8%). This reflects which studies reported on which basis (a data-availability artifact across studies), not a within-sample dry/wet conversion, so it does not by itself indicate the dry basis runs higher for seaweed. (The +77% dry-weight figure in Table 4.1 is the seafood row, not seaweed.)
• Cooking/preparation: traditional washing and soaking of Hizikia (hijiki) can reduce total As by up to ~60% (citing EFSA 2009); cooking rice in high water volumes can reduce inAs.
• Data-quality caveats specific to algae: extraction efficiencies are often <80% (one study spanned 20.7–97% across species), which may underestimate total As; varying LODs and combined dry/wet-weight reporting add uncertainty; the authors stress speciation-based, study-quality-weighted, population-tailored intake calculations.

Methods (brief)

Comprehensive literature search (PubMed, Scopus, Toxline, Agricola) for studies measuring speciated arsenic in food intended for human consumption, 1995–July 2013; 134 studies retained after excluding anthropogenically-contaminated-site and laboratory-uptake studies. Each food was categorised into food groups; for each group the authors computed mean (mean-of-study-means or arithmetic mean of raw data, weighted by sample size), minimum and maximum for inAs³⁺, inAs⁵⁺, total inAs, MMA, DMA, and total As. Values below LOD were substituted with half the LOD (with sensitivity analyses using 0, minimum LOD, and minimum detected value); dry- and wet-weight values were combined in the main analysis (with dry-vs-wet sensitivity analysis). No primary sampling or measurement was performed. Funded by the Electric Power Research Institute (EPRI); authors affiliated with Gradient (a scientific consulting firm) — disclosure recorded for transparency.

Implications

• Certification (HMTc): a B-tier literature-wide context anchor for the Category 6 seaweed-kelp-foods row (iAs/tAs platform). It establishes that across the published literature (n=312 inAs values) seaweed/algae carries the highest central-tendency inorganic arsenic of any food group (mean 11,000 µg/kg, range <14–120,000), with a wide spread driven by species (hijiki/Sargassum and Laminaria high) and high non-detect fractions in low-iAs species. (The row’s native dry, as-sold basis rests on the certification methodology and the primary occurrence sources, which report dry weight — not on this meta-analysis’s dry/wet sensitivity split, which for seaweed is a cross-study data-availability artifact.) Because these are meta-analytic aggregates of the underlying primary studies (several of which are ingested separately in this folder), they must be treated as context only and excluded from the occurrence-extraction/pooling pass to avoid double-counting. The EPRI funding / consulting authorship is noted for defensibility but does not change the tier.
• Courses: a useful overview of why inAs (not total As) and study quality, LOD handling, and dry/wet basis matter when estimating dietary arsenic exposure.

Verification notes

• raw_handle MFK_lynch2014 from the PDF filename; raw_path under “raw/Manual Fetch Kimi /June 8 Inorganic Arsenic Seaweed/“. DOI 10.1016/j.scitotenv.2014.07.032 confirmed on the article header.
• Evidence tier B, source_type review: this is a literature meta-analysis (secondary). Per Part 13, its claims are B-tier attributed context/leads. sample_n is null (no primary samples); the per-group n values (e.g., seaweed inAs n=312) are counts of compiled literature data points, recorded in Key numbers, and are explicitly flagged as not-poolable to prevent double-counting the primary studies.
• Speciation: total As, total inorganic As, and the inAs³⁺/inAs⁵⁺ split are reported and kept distinct (lifted to frontmatter as tAs, iAs). MMA and DMA are reported and summarised but NOT lifted to frontmatter (not HMI analyte-vocabulary tokens); recording them as iAs would be a speciation error.
• Units/basis: all figures reproduced as the review reports them (µg/kg; the main analysis combines dry and wet weight, with the dry-weight sensitivity result noted). Not for pooling.
• jurisdictions [US]: although the compiled data are global, the analysis is US-framed (US dietary-intake/risk-assessment purpose, US EPA IRIS slope factor, US FDA/NRC context, US authors, EPRI funding); US is its analytical lens, parallel to other regionally-framed reviews. The seaweed data origins themselves are global (Denmark, China, Bangladesh, etc.), noted here.
• Brand firewall: not engaged. The EPRI funding source and the Gradient affiliation are disclosure facts, not brand-by-contamination attributions; no food brands are named.
• matrices [edible-seaweed, macroalgae, dry-weight] and products [seaweed-kelp-foods]: this is a multi-food review, but it is ingested here for its seaweed content and routed to the seaweed row; its rice/seafood/infant-food figures are captured in Key numbers as cross-food context but the routing is intentionally scoped to seaweed (a future session may enhance routing to other product rows if desired). kelp omitted from matrices because the review reports the aggregate “seaweed/algae” group, not Laminaria specifically.
• Audit subagent (2026-06-08, fresh-context) returned REVISE; all ~30 other values, speciation labels (tAs/iAs; MMA/DMA not mislabeled), slugs, and both firewalls verified clean, and the not-poolable / B-tier / [US] framing confirmed appropriate. One ❌ applied: the dry-weight sensitivity figure had been transposed from the wrong Table 4.1 row — the +77% is the seafood row; the actual seaweed dry-weight-only value is −38% (a cross-study data-availability artifact). Corrected the figure and removed the dependent (and reversed) inference that the dry basis “trends higher” for seaweed. (The error was seeded by the ingest instruction; the auditor caught it against the source.)
• Cross-source note: this meta-analysis aggregates many primary studies, several held in this same folder (e.g., Almela, Rose, Llorente-Mirandes, Narukawa); any cross-source reconciliation belongs to the Part 9 synthesis pass, not this page.

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.