EFSA Scientific Committee 2015 — Risk-benefit assessment of fish/seafood consumption versus methylmercury

This EFSA Scientific Committee statement is the European Union’s primary risk-benefit synthesis weighing the nutritional benefits of fish/seafood consumption (n-3 long-chain polyunsaturated fatty acids, LCPUFA) against exposure to methylmercury (MeHg) for the European population. EFSA received the mandate from the European Commission as Question No EFSA-Q-2014-00665 and adopted the statement on 19 December 2014. The Scientific Committee built on two prior EFSA outputs as starting points: the EFSA CONTAM Panel (2012) opinion on inorganic mercury and methylmercury in food, which had set the operative tolerable weekly intake (TWI) of 1.3 µg/kg body weight per week for methylmercury based on prenatal neurodevelopmental toxicity (NOEL ~11 mg/kg maternal hair after adjustment for n-3 LCPUFA confounding in the Seychelles cohort, with an uncertainty factor of 6.4), and the EFSA NDA Panel (2014) opinion on the health benefits of seafood consumption in relation to methylmercury risks, which concluded that consumption of about 1–2 servings up to 3–4 servings of seafood per week during pregnancy and adulthood is associated with beneficial effects on functional neurodevelopment in children and on coronary heart disease (CHD) mortality in adults. The contribution of this 2015 statement is a scenario-based calculation, executed for each population group across 17 EU Member States, of how many fish/seafood servings per week would be required to reach the TWI for methylmercury and the dietary reference value (DRV) for n-3 LCPUFA, given that group’s reported pattern of contributing species and serving sizes. The Committee found that toddlers (1–<3 years) and other children (3–<10 years) are the age classes that reach the TWI at the fewest servings per week (typically 2–3 servings/week, and in some cases below 1 serving/week where high-MeHg predatory species dominate consumption), and concluded that no single Europe-wide fish-consumption recommendation is possible — each country must assess its own pattern of fish species consumed, limit intake of species high in methylmercury (notably swordfish, tuna, pike, hake, bass, bream, lobster, and Lophiiformes/monkfish) to protect women of child-bearing age and young children, and prefer species low in methylmercury (notably salmon/trout, herring, mackerel, cod, plaice, sole, squid, fish products, and whitefish) where the benefits of fish consumption can be obtained at substantially higher servings per week.

Key numbers

Mandate, basis, and scope (Sections 1–2):

• Mandate received from the European Commission as EFSA-Q-2014-00665 (Article 29(1)(a), Regulation (EC) No 178/2002); statement adopted by the Scientific Committee on 19 December 2014.
• Operative TWI for methylmercury: 1.3 µg/kg bw/week, expressed as Hg, established by the EFSA CONTAM Panel (2012) on prenatal neurodevelopmental toxicity (Seychelles cohort reanalysis adjusting for prenatal maternal blood n-3 LCPUFA, identifying an apparent NOEL of ~11 mg/kg maternal hair; uncertainty factor 6.4 covering kinetic and dynamic variability from maternal hair to maternal dietary intake).
• Adequate intakes (AI) and DRVs for n-3 LCPUFA (EPA+DHA) used in this statement: 250 mg/day for adults (cardiovascular endpoint); 100 mg/day DHA for infants/toddlers 6–24 months; 125 mg EPA+DHA/day used for toddlers in absence of a specific value (with 250 mg/day for older children, adolescents and elderly extrapolated from the adult value); 350 mg/day for women of child-bearing age (250 mg EPA+DHA plus 100–200 mg preformed DHA during pregnancy/lactation).
• Assumption for fish species composition: methylmercury comprises ~80–100 % of total mercury in fish meat, fish products, fish offal and unspecified fish and seafood, and ~50–80 % in seafood other than fish; the CONTAM (2012) exposure assessment assumed 100 % methylmercury in fish meat, fish products, fish offal and unspecified fish/seafood, and 80 % in crustaceans, molluscs and amphibians (other foods assumed inorganic Hg only).
• Statement explicitly does NOT address inorganic mercury, dioxins/dioxin-like compounds, or other contaminants in fish/seafood; the Scientific Committee refers to EFSA CONTAM (2005) for those.

Per-species methylmercury and n-3 LCPUFA inputs (Appendix C; MeHg µg/kg, n-3 LCPUFA mg/100 g):

Swordfish has the highest MeHg level among all species in Appendix C at 1,212 µg/kg, an order of magnitude above the species-group mean for fish meat. Salmon/trout (33 µg/kg) and herring (36 µg/kg) are the lowest among species reported, while also providing 1,815 and 2,482 mg/100 g of n-3 LCPUFA respectively, making them the most favourable species in this risk-benefit balance. Tuna combines high MeHg (290 µg/kg) with high n-3 LCPUFA (2,806 mg/100 g), so it functions as a high-leverage contributor on both sides of the balance.

Servings per week to reach TWI for MeHg vs DRV for n-3 LCPUFA (Appendix D, selected entries; servings/week):

Population groups exceeding the TWI for methylmercury at the fewest servings per week (Section 3.2 and Figure 1):

• Toddlers (1–<3 years) and other children (3–<10 years) are the age classes that reach the TWI at the fewest servings; in Italy “other children” reach the TWI at 0.5 servings/week (the lowest figure in Appendix D).
• In a few cases — including Italian other children, Italian adolescents, Italian adults, Italian elderly, Italian women of child-bearing age, and Spanish other children (enKid) — the TWI is reached after less than or around 1 serving per week, driven by high swordfish/tuna/bream/bass consumption.
• In most other cases for toddlers and other children the TWI is reached at 2–3 servings/week; in most cases the TWI for mercury is reached with 2–4 servings/week.

P95 methylmercury exposure by country and age group (Appendix E, µg/kg bw/week, middle bound):

Italy “other children” P95 exposure of 4.96 µg/kg bw/week is 3.8× the TWI; Italy adolescents (5.05), Italy women of child-bearing age (5.03), and Spain other children enKid (4.69) all sit at ~3.5–3.9× the TWI. The Spanish enKid adolescents (3.45) and Italian P95 adults (3.04), Spanish Aesan adults (2.98), and Spanish Aesan-Fiab and Aesan women of child-bearing age (3.08, 3.05) all exceed 2× the TWI. The driving species are swordfish (in Italy and Spain), hake (in Spain), tuna (across both countries), and pike (in Finland and parts of Germany).

Serving sizes (Section 3.2 and Appendix E): Reported serving sizes for “fish and other seafood” across population groups ranged from 23 g (lowest, Finland toddlers) to 135 g (highest); the lowest serving sizes were for toddlers (23–100 g) and other children (38–82 g). In the majority of cases reported serving sizes were below 100 g. Several serving-size values flagged in Appendix E in red are based on fewer than 10 eating events and are therefore highly uncertain.

Risk-benefit framing (Section 3.2 and Figure 1):

• General population-level conclusion: in the range of 1–4 servings/week, the benefits of fish consumption outweigh the risks, whatever the respective contribution to this net effect of beneficial and harmful fish components may be.
• For some population groups (those above the dotted line in Figure 1), the TWI for methylmercury is reached at a lower number of servings per week than needed to reach the DRV for n-3 LCPUFA — i.e., benefits cannot be attained without exceeding the TWI unless the species mix is shifted toward lower-MeHg species.
• Several groups reach the TWI and the DRV at approximately the same number of servings per week.
• An uncertainty factor of 6.4 is built into the TWI derivation (covering toxicokinetic extrapolation from mercury in hair to mercury intake); no comparable uncertainty factor is used in setting the DRV for n-3 LCPUFA.

Methods (brief)

• Risk-benefit framework: EFSA Scientific Committee 2010 guidance on human health risk-benefit assessment of food (EFSA Journal 2010;8(7):1673) — tiered approach in which risk and benefit elements are characterised separately before being weighed against each other.
• Methylmercury occurrence data: drawn from the EFSA CONTAM Panel (2012) opinion (EFSA Journal 2012;10(12):2985), which assembled ~60,000 mercury results from 20 European countries (mainly 2004–2011), of which 36.8 % were for the food group “Fish and other seafood”. Per-species mean MeHg concentrations (Appendix C) calculated by applying the speciation assumption (100 % MeHg in fish meat/products/offal/unspecified, 80 % MeHg in crustaceans/molluscs/amphibians) to total Hg measurements; no fresh analytical work was performed by the Scientific Committee for this statement.
• n-3 LCPUFA content data: drawn primarily from the EFSA NDA Panel (2014) opinion (EFSA Journal 2014;12(7):3761). For swordfish, n-3 LCPUFA content was calculated from the EU-funded CALIPSO project (Leblanc 2006) using the same C20:5(n-3) + C22:6(n-3) formula as the NDA Panel. For whitefish, the Finnish food composition database Fineli was used (750 mg/100 g).
• Consumption data: EFSA Comprehensive European Food Consumption Database. Only population groups composed of at least 20 individuals were considered (Spanish toddlers, based on 17 individuals, were excluded). The 26 chronic dietary surveys across 17 EU Member States in Appendix B include 2-day 24-hour dietary recalls (Bulgaria, Czech Republic, Germany National Nutrition Survey II, Spain enKid/NUT-INK05/AESAN, Finland FINDIET 2007, Latvia, Sweden NFA) and 2-day to 7-day dietary records (Belgium, Cyprus, Germany DONALD 2006/2007/2008, Spain AESAN FIAB, Finland DIPP/STRIP, France INCA2, Italy INRAN SCAI 2005/06).
• Scenario construction: rather than directly translating daily-recall data to weekly intake (which would overestimate the upper tail), the Scientific Committee constructed exposure scenarios from the composition of the main contributing fish/seafood species (those individually contributing ≥5 % of total MeHg exposure for that group) and the reported serving sizes for that group. For each population group, a single weighted “fish/seafood serving” was computed as Σ (% contribution × serving size) / Σ (% contribution). Exposure to MeHg and intake of n-3 LCPUFA from one serving were then computed by multiplying that weighted serving by the per-species MeHg and n-3 LCPUFA concentrations in Appendix C. The number of servings per week to reach the TWI (1.3 µg/kg bw/week) and the DRV for n-3 LCPUFA was derived from these per-serving values.
• Population groups: age classes per the EFSA Comprehensive Database (Appendix A) — infants (0–11 months), toddlers (12–35 months), other children (36 months–9 years), adolescents (10–17 years), adults (18–64 years), elderly (65–74 years), very elderly (≥75 years), and women of child-bearing age (18–45 years). Fish consumption data of women of child-bearing age were used as a proxy for women in the year preceding pregnancy and during pregnancy.
• Worked example (Section 2.2, Belgian “other children”): serving size for the group was 89.8 g, computed as (0.37 × 105.6 + 0.28 × 78.3 + 0.14 × 84.9 + 0.07 × 62.2) / (0.37 + 0.28 + 0.14 + 0.07); per-serving MeHg exposure was 6.48 + 2.05 + 0.45 + 0.14 = 9.12 µg/week, divided by 7 to get the per-day equivalent and by the average bodyweight 17.88 kg gave 0.6 µg/kg bw/week; one serving per week delivered 91.3 mg of n-3 LCPUFA. The TWI for methylmercury was therefore reached at 2.2 servings/week, and the DRV for n-3 LCPUFA at 2.7 servings/week.

Uncertainties (per Section 3.3)

• The MeHg concentrations used here are inherited unchanged from the CONTAM (2012) opinion; some fish/seafood commodities reported in consumption surveys were not well characterised in the occurrence dataset and assumptions had to be made about their MeHg content.
• The 2014 NDA opinion attributes the net benefit of fish consumption to “eating fish” rather than isolating individual nutrients (n-3 LCPUFA, selenium, iodine, vitamin D); the present statement uses n-3 LCPUFA intake as a tractable marker of benefit but acknowledges this is a partial proxy.
• Per-species serving sizes and contribution percentages were computed group-by-group from the consumption database; in several cases (entries flagged in red in Appendix E) the calculation relied on fewer than 10 eating events and is therefore highly uncertain, with the direction of the resulting bias unknown.
• The category “Redfish” reported by several Member States was matched to “Ocean Perch” for n-3 LCPUFA content, an ad-hoc extrapolation whose net bias direction is unknown.
• “Pike”, “Swordfish” and “Redfish” are not referenced in the FoodEx classification used in CONTAM (2012) and had to be created ad hoc; European-average serving sizes for these species were averaged across the individual eating events that mentioned them.
• “Norway Lobster” reported by Spain was matched to “Lobster (Homarus vulgaris)” for n-3 LCPUFA content, another ad-hoc extrapolation.
• Daily-to-weekly extrapolation: dietary surveys conducted over 2-day or 3-day windows tend to overestimate the upper tail of weekly intake when scaled by 7; the scenario approach used here is meant to mitigate this but does not eliminate it.
• Cross-sectional consumption data are not paired with clinical/physiological endpoints in this population, so the risk-benefit balance is necessarily reconstructed indirectly from prior NDA Panel (2014) findings rather than measured directly.

Implications

This statement contributes the operative EU-level synthesis of how methylmercury intake from fish/seafood interacts with the n-3 LCPUFA benefits of fish/seafood consumption across European population groups. For the wiki it provides:

• Per-species MeHg occurrence reference values (Appendix C) that are derivable to the canonical fish/seafood ingredient and product pages, anchored on the CONTAM (2012) Hg dataset and the published 100 %/80 % MeHg speciation assumptions.
• Quantitative servings-to-TWI estimates by country and age group (Appendix D) that can inform discussion on [[metals/mercury-methyl]], [[products/fish-marine-predatory]], and the broader fish-consumption discussion on [[products/fresh-fish]] and [[products/seafood]].
• The operative EU TWI for methylmercury (1.3 µg/kg bw/week) cited via this statement and traceable to its CONTAM (2012) derivation; the wiki regulation page [[regulations/efsa-methylmercury-twi]] is the canonical record of that threshold.
• A risk-benefit framing that demonstrates why simple “fish is good” or “fish is bad” messaging fails: in the range of 1–4 servings/week the population-level balance favours benefit, but at the species level swordfish, tuna, pike, hake, bream and bass are sufficient on their own to exceed the TWI at modest servings, while salmon/trout, herring, mackerel and cod allow the n-3 LCPUFA DRV to be met well below the TWI. The wiki ingredient/product pages can cite this contrast directly when discussing fish species selection.

Wiki pages this source may touch

• mercury-methyl — operative EU methylmercury TWI (1.3 µg/kg bw/week), per-species MeHg occurrence values in fish/seafood, population groups exceeding the TWI.
• mercury-total — speciation assumption (100 % MeHg in fish meat/products/offal/unspecified, 80 % MeHg in crustaceans/molluscs/amphibians) used in deriving MeHg from tHg measurements.
• mercury — overall mercury narrative for fish/seafood.
• fish — per-species MeHg values for the most-consumed European fish species; the operative basis for the EU fish-consumption framing.
• seafood — fish + crustaceans + molluscs MeHg/n-3 LCPUFA values.
• shellfish — lobster, squid, Lophiiformes MeHg values.
• freshwater-fish — perch, pike, carp MeHg values from the European dataset.
• swordfish — highest-MeHg species in Appendix C (1,212 µg/kg).
• shark — predatory-fish context (per the wider CONTAM 2012 evidence base; this statement does not assign a per-species MeHg value to shark).
• canned-tuna — tuna (290 µg/kg MeHg, 2,806 mg/100 g n-3 LCPUFA) as a high-leverage species on both sides of the balance.
• fresh-fish — broad fresh-fish category; per-species MeHg/n-3 LCPUFA values applicable.
• canned-fish — canned fish (notably canned tuna) MeHg occurrence and consumption-pattern context.
• seafood — umbrella seafood product page.
• shellfish — lobster, squid, monkfish.
• fish-marine-predatory — swordfish, tuna, pike, bream, bass MeHg risk.
• fish-marine-non-predatory — herring, mackerel, salmon, cod MeHg/n-3 LCPUFA favourable balance.
• fish-freshwater — perch, pike, carp MeHg risk in freshwater species.
• efsa-methylmercury-twi — operative EU TWI of 1.3 µg/kg bw/week, derivation and uncertainty factor 6.4.
• jecfa-methylmercury-ptwi — JECFA PTWI (1.6 µg/kg bw/week) referenced in the same evidence base.
• eu-reg-2022-617-mercury-fish — downstream maximum-levels regulation that operationalises mercury limits in fish/seafood at the EU level.

Verification notes

• Cite-key disambiguation: existing efsa2015-nickel-food-water (EFSA CONTAM Panel 2015 Ni opinion, DOI 10.2903/j.efsa.2015.4002) is a different EFSA panel and different topic; this statement uses cite-key efsa-sc2015-fish-seafood-mehg-risk-benefit to make the Scientific Committee provenance explicit and to differentiate from CONTAM and NDA panel outputs in the same year.
• Speciation field choice: the paper explicitly addresses methylmercury throughout, with total mercury as the underlying analytical measurement converted via the 100 %/80 % assumption documented in CONTAM (2012). Both MeHg and tHg are included in metals: because the per-species table (Appendix C) reports MeHg µg/kg, while the additional-information section discusses tHg in fish in the broader CONTAM (2012) context.
• Ingredient slug selection: restricted to slugs present in the current taxonomy snapshot (fish, seafood, shellfish, freshwater-fish, swordfish, shark, canned-tuna). Per-species values for cod, hake, bream, bass, perch, pike, mackerel, herring, salmon, lobster, sole, plaice, squid, redfish, whitefish, monkfish are documented in the Key numbers section but do not have their own ingredient pages in the current taxonomy; the routing layer fans the broader fish / seafood / freshwater-fish slugs to siblings as needed.
• Product slug selection: uses both the broad fresh-fish/canned-fish/seafood/shellfish umbrella slugs and the form-specific fish-marine-predatory/fish-marine-non-predatory/fish-freshwater slugs because the per-species data clearly route to all three predator/non-predator/freshwater subgroups.
• Sample size: sample_n: null because this is a scenario-based risk-benefit assessment, not a primary occurrence study; the occurrence and consumption datasets it integrates have their own sample sizes documented in CONTAM (2012) and the EFSA Comprehensive Food Consumption Database respectively.
• Brand firewall: no commercial brand names appear in this paper; all species are identified by common species name or by the FoodEx grouping (Fish meat, Fish products). Cohort/study names (DIPP, STRIP, NUT INK05, Aesan-Fiab, enKid, NFA, INCA2, INRAN SCAI, FINDIET, DONALD, NUTRICHILD, SISP04, etc.) are dietary survey identifiers, not commercial brands, and are retained as method context.
• Audit subagent (2026-06-03, fresh-context general-purpose) verdict PROMOTE. All five checks ✅ except Check 1 ⚠️ on the Belgian worked-example per-contributor values (6.48/2.05/0.45/0.14 µg/week). Verified against source as a false positive: PDF page 13’s contributor table prints rounded values 6.5/2.1/0.5/0.1, but PDF page 14 explicitly writes the per-serving MeHg computation as [(6.48+2.05+0.45+0.14)/(0.86×17.88)] = 0.6 µg/kg bw/week. The wiki page reproduces the page-14 unrounded equation verbatim. No correction applied; subagent saw the p. 13 table but missed the p. 14 explicit equation.

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.