Mahaffey et al. 1975 — FDA Total Diet Study: Heavy Metal Exposure from Foods

This 1975 paper presents the first multi-year US dietary monitoring data for six trace elements (Pb, Cd, total Hg, total As reported as As₂O₃, Zn, Se) derived from the FDA Total Diet Study (Market Basket Survey). The protocol composites 117 foods into 12 food classes prepared as consumed, and uses 30 market baskets per year from four US geographic regions as the basis for national-average daily intake estimates. The authors report (i) detection frequencies and mean concentrations by food class for FY 1973 (Tables 5, 6), (ii) yearly total intake trends 1968–1974 (Table 9), and (iii) FY 1973 intakes against WHO/FAO Provisional Tolerable Weekly Intakes (Table 10). The headline finding: average US adult intakes of Pb, Cd, and tHg all fell below WHO/FAO tolerable levels, but cadmium intake reached 71.7–89.7% of the provisional tolerable, prompting the authors’ explicit warning that “further increases in the cadmium content of foods should be avoided.” Lead intake was acknowledged as likely underestimated because many composites were at or below the analytical detection limit. No WHO/FAO tolerable intake had been established for As or Se at the time of writing. The paper is the methodological and baseline reference for the US market-basket heavy metals surveillance approach.

Key numbers

All concentrations below are in mg/kg (ppm) wet weight, as-consumed prepared composites. Listings in Table 5 are presented as detection % (mean ppm across the 30 market baskets, including non-detects as zeros for the purpose of averaging traces).

Estimated total daily intakes, FY 1973 (Table 6 totals; Tables 9–10)

• Lead (Pb): 60.4 µg/day total, equivalent to 0.87 µg/kg body weight/day, or 14.1% of the WHO/FAO Provisional Tolerable Weekly Intake (PTWI; 3.0 mg/person/week ≈ 429 µg/day). The authors caution this is likely an underestimate because Pb in many composites is at or below the method detection limit; Table 7 (re-tabulated from Kolbye et al. 1974) gives 57 µg/day when zeros and trace amounts are excluded, 159 µg/day when traces are assigned 0.09 ppm, and 233 µg/day when both traces (0.09 ppm) and zeros (0.05 ppm) are imputed at the detection limit. The 1965–1970 average from earlier work was 63 µg/day.
• Cadmium (Cd): 51.2 µg/day total in FY 1973 (Table 6), equivalent to 0.73 µg/kg body wt/day, or 71.7–89.7% of the WHO/FAO PTWI (0.4–0.5 mg/person/week ≈ 57.1–71.4 µg/day). FY 1974 intake fell to 34 µg/day (Table 9). Top contributors: grain and cereal products 11.66 µg/day (22.8% of Cd intake), fruits 9.38 µg/day (18.3%), beverages 6.49 µg/day (12.7%), legume vegetables 3.18 µg/day, dairy 3.94 µg/day.
• Mercury (total Hg): 2.89 µg/day total in FY 1973 (Table 6), equivalent to 0.041 µg/kg body wt/day, or 6.7% of the WHO/FAO PTWI for total Hg (0.3 mg/person/week ≈ 42.9 µg/day). Effectively 100% of dietary Hg came from the meat–fish–poultry composite, dominated by fish. All other food-class composites returned no detectable Hg by flameless AAS; the authors note neutron-activation analysis would be required to detect Hg in non-fish matrices.
• Arsenic (total As, reported as As₂O₃): 10.1 µg/day (FY 1973, Table 6), 0.15 µg/kg body wt/day. Top contributors: meat–fish–poultry 5.64 µg/day (55.6%), dairy 2.34 µg/day (23.1%), grain and cereal 1.35 µg/day (13.7%); together these three classes account for 92% of intake. WHO/FAO PTWI not established at the time of writing.
• Zinc (Zn): 17.9 mg/day in FY 1973 (Table 6 totals 17,918 µg). Top contributors by mass: meat–fish–poultry 6.66 mg/day (37.2%), dairy 3.84 mg/day (21.4%), grain and cereal 3.37 mg/day (18.8%). Zn was detected in all 12 food classes.
• Selenium (Se): 149.7 µg/day in FY 1973 (Table 6). Top contributors: grain and cereal 92.5 µg/day (61.8%), meat–fish–poultry 56.3 µg/day (37.6%); these two classes account for >99% of dietary Se. WHO/FAO PTWI not established for Se at the time of writing.

Distribution of major dietary sources (Tables 3, 4)

• Cadmium: generalised distribution; highest concentrations in grain and cereal products. Specific high-Cd composites cited in body text: shellfish and spinach-type leafy vegetables.
• Lead: generalised distribution; vegetables and garden fruits together account for 78% of total Pb intake despite comprising only 23% of the diet by weight. Table 3 lists the contributing pair as “Vegetables, all types combined” 44% and “Fruits and garden fruits” 34%.
• Zinc: dairy 21%, meat-fish-poultry 37%, grains/cereals 19% (combined 77%).
• Arsenic (as As₂O₃): dairy 31%, meat-fish-poultry 56%, grain/cereal 14% (combined ≈100%).
• Selenium: meat-fish-poultry 38%, grain/cereal 62% (combined >99%).
• Mercury (total): meat-fish-poultry 100%.

Detection frequency and mean concentration by food class, FY 1973 (Table 5)

Listed as detection % (mean ppm wet weight including zeros). Note that traces below the quantitation limit count as positive occurrences but contribute 0 to the mean — the authors flag this as the main source of Pb underestimation. Zinc and Selenium columns omitted from the table below; see body text and Table 6 for those values.

Yearly total daily intake, 1968–1974 (Table 9)

Daily intake in µg/day except Zn (mg/day). “ND” = not determined; Pb/Zn/Se measured from 1972 onward only.

c = 1965–1970 average for As and Cd was 63 µg/day (combined estimate from earlier years); 1971–1972 As intake not calculated at time of writing. 1974 values are flagged in the original table as preliminary/estimated.

The authors state these year-over-year data “indicate no trends of increasing or decreasing metal exposure in the diet” over the period studied.

Methods (brief)

FDA Total Diet Study (Market Basket Survey) protocol. 117 individual foods, based on the FY 1965 USDA dietary survey of 15- to 20-year-old males (the highest-calorie consumer group in the population), assembled into 12 food classes. 30 market baskets purchased per year from retail stores in four US geographic regions (Southeast, Northeast, Central, Western). Each food prepared as normally consumed (vegetables boiled if usually boiled, meats cooked, etc.) using a dietetic kitchen in Washington, DC. The 12 food classes were then composited and analysed; individual foods within a class were not analysed unless a class composite returned an unusually high concentration. Quantities of each food in the composite are based on the USDA survey data. The paper does not give analytical detail beyond brief notes: Pb in many composites was at or below the method detection limit (the resulting underestimation problem is acknowledged but not quantified). Hg was determined by flameless atomic absorption spectrometry; non-fish matrices required neutron activation analysis to detect any Hg (citing Simpson et al. 1974). The authors cite Kolbye et al. 1974 (EHP 7: 65) for the prior treatment of analytical assumptions on Pb. Speciation was not performed: As is reported as the total-element oxide (As₂O₃) rather than separately as inorganic vs organic, and Hg is reported as total Hg rather than methylmercury. The reference for the underlying compliance protocol is FDA Compliance Program Guidance Manual 7320.08, Total Diet Studies—Adult, 1974.

Implications

Certification: This is the historical baseline reference for US dietary heavy metal exposure prior to the modern regulatory era. It documents that as early as FY 1973, average dietary cadmium intake in US 15–20-year-old males was 72–90% of the WHO/FAO Provisional Tolerable Weekly Intake — well before Cd in plant foods was a regulatory priority. For HMTc threshold work, the paper contributes occurrence-pattern data (Pb dominated by vegetables and garden fruits; Cd dominated by grains, fruits, and beverages; Hg dominated by fish) that align with modern food-class attribution patterns; it does not contribute occurrence data at ingredient-level resolution because the composite-class methodology pools heterogeneous foods.

Courses: The 1975 Total Diet Study methodology is the origin of the market-basket approach FDA continues to use. Useful as the entry point for explaining national dietary surveillance infrastructure to brand-side educational audiences, and for showing how trace-handling assumptions (zeros, traces below the quantitation limit) can swing total intake estimates by 4× for a single metal (Pb: 57 → 233 µg/day).

App: Data are at composite-class resolution and 50+ years old; not usable as occurrence inputs for per-ingredient contamination_profile blocks. The qualitative finding that grains/cereals dominate Cd intake and that fruits and vegetables dominate Pb intake remains consistent with modern ingredient-level findings, but the numerical values themselves are outdated.

Verification notes

• 2026-05-18 (Claude session, merge-enhance): Re-read full PDF (7 pages, all tables) against the prior version of this page. Frontmatter corrections: (1) raw_handle changed from the generic legacy value papers-cube to the current convention PCMF_heavy-metal-exposure-from-foods; (2) raw_sha256 added (75cea057…); (3) metals expanded to include Se (the paper measures and tabulates dietary selenium, omitted from prior frontmatter); (4) ingredients populated with the eleven food-class equivalents the paper measures (milk-and-dairy, meat-and-poultry, fish, cereals, potatoes, leafy-vegetables, legumes, root-vegetables, vegetables, fruits, beverages); (5) matrices reduced from [dietary-intake, composite-food-basket] to [dietary-intake] — the second term is not in the controlled vocabulary in docs/gpt-collaboration/system-prompt.md and was an invented matrix slug.
• 2026-05-18 (Claude session): Prior version of Key numbers asserted “60 µg/day (Method: zeros treated as zeros), ranging up to 233 µg/day under alternative censoring assumptions.” This was an incorrect attribution. Verification against Table 7: the 60.4 µg/day figure (Table 6) is the standard estimate with trace values contributing 0 to the mean but counted as occurrences; the “zeros treated as zeros” (i.e., zeros and traces excluded from the calculation entirely) assumption yields 57 µg/day per Table 7. The full range 57–233 µg/day reflects four different trace-handling assumptions from the Kolbye et al. 1974 re-tabulation cited as ref [3]. Section rewritten to attribute each value to its actual source row in Table 7.
• 2026-05-18 (Claude session): Prior version omitted the WHO/FAO % of tolerable daily intake numbers in Table 10, which are the headline result the paper’s abstract emphasises. Added: Pb at 14.1% of WHO/FAO PTWI, total Hg at 6.7%, and Cd at 71.7–89.7% (range reflecting WHO/FAO’s 0.4–0.5 mg/person/week range for Cd). The Cd range explains the paper’s explicit warning that “further increases in the cadmium content of foods should be avoided.”
• 2026-05-18 (Claude session): Table 5 figures corrected from prior version’s 3-significant-figure rounding back to the source’s 4-significant-figure precision (0.003 → 0.0033 for several entries; 0.009 → 0.0093 for the meat-fish-poultry Cd value). Three Table 5 rows previously omitted (Oils/fats/shortenings, Sugars/adjuncts, Beverages) added back; the table is now complete for all 12 food classes for Pb, Cd, tHg, and As. Yearly trend table (Table 9) added; this is the year-over-year data the paper relies on to claim “no trends of increasing or decreasing metal exposure.”
• 2026-05-18 (Claude session): Speciation is correct as labeled tHg and tAs: the paper explicitly reports total Hg (not MeHg) and total As expressed as As₂O₃ (not iAs); the footnote on page 1 acknowledges that “the term ‘heavy metals’ is used in a colloquial sense and includes the toxic nonmetals arsenic and selenium.” No iAs or MeHg speciation should be inferred from this paper.
• 2026-05-18 (Claude session): Dead-wikilink cleanup. Prior version listed [[health/exposure-routes]] and [[testing/total-diet-study-methodology]] under “Wiki pages updated on ingest”; neither page exists. Replaced with links to extant metal pages plus a forward reference under Implications. The “Wiki pages updated on ingest” header is preserved per template convention even where the list shortens.
• DOI status: doi: null, no_doi_assigned: true retained from prior version. Environmental Health Perspectives back-issue retrospective DOIs (10.1289/ehp. form) are plausible for this article but were not verified against the publisher record in this session. If the paper acquires a verified DOI, both fields should be updated together.
• 2026-05-18 (audit subagent flagged ❌ — verified against PDF p. 64 Table 3 and narrative — applied): First-pass merge-enhance erroneously stated under the Lead bullet that “dairy = 23% of intake.” No such figure exists in the paper. Table 3’s Pb row lists “Vegetables, all types combined 44” and “Fruits and garden fruits 34,” totaling the 78% the abstract reports. The 23% number that appears in the source’s Pb narrative refers to “23% of the total diet by weight” (vegetables and fruits as a share of total dietary mass), not to a dairy contribution. The 21% dairy figure that does exist in Table 3 is for Zn, not Pb. Lead bullet rewritten to quote Table 3’s actual Pb percentages and remove the fabricated dairy figure.
• 2026-05-18 (audit subagent ⚠️ — verified — left as-noted): Subagent observed that selenium is measured and tabulated in the paper but the page lists no [[metals/selenium]] link under “Wiki pages updated on ingest.” Confirmed by ls wiki/metals/: no selenium page exists in the current wiki (the HMTc analyte panel is Pb / tAs / Cd / MeHg / tHg / iAs / Ni / Al / Cr-VI / Sn; Se is not an HMTc analyte and has not been promoted to its own metal page). Per Part 10, freq-1 metal mentions do not warrant page creation. Selenium data are preserved in the source page’s Key numbers; no downstream page exists to link to.

Wiki pages updated on ingest

• lead
• cadmium
• mercury-total
• arsenic-total
• zinc

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.