US House Subcommittee 2021 — Baby Foods Are Tainted with Dangerous Levels of Arsenic, Lead, Cadmium, and Mercury

The U.S. House Subcommittee on Economic and Consumer Policy compiled internal testing records subpoenaed from seven major US baby-food manufacturers between November 2019 and February 2021 and reported sample-level inorganic arsenic, total arsenic, lead, cadmium, and mercury concentrations across ingredient pre-shipment tests and finished-product tests. The staff report aggregates manufacturer-provided test results for products marketed for infants and young children, identifies internal corporate threshold values that exceed every existing US and international regulatory standard for related products (drinking water, infant formula, juice, candy), documents a confidential August 2019 Hain presentation to FDA showing that finished-product inorganic arsenic levels run 28-93% higher than ingredient testing alone predicts, and recommends mandatory finished-product testing, mandatory labeling, voluntary phase-out of high-contamination ingredients (notably rice), uniform FDA action levels across all infant and toddler foods, and parental avoidance guidance. The report is a public-record Congressional oversight document; the manufacturers named are subjects of the investigation per Part 12 Exception 1.

Key numbers

The report presents ingredient-level and finished-product-level test concentrations, internal corporate threshold values, and a synthesis of cited prior-art exposure thresholds. Values below are organized by analyte; the ingredient-level results are the primary scientific content. Manufacturer attribution is preserved only where the regulatory-event context requires it (which company refused to cooperate, which company’s confidential FDA presentation is the subject of the report’s central finding). Per-manufacturer ranking tables from the report are not reproduced.

Inorganic arsenic (iAs) — finished products and ingredients. Finished baby-food products tested at concentrations up to 180 ppb iAs, with one major cooperating manufacturer reporting an average of 59.54 ppb iAs across 113 finished products and 78% (89 of 113) of those finished products at 9 ppb iAs or higher; over 25% of that manufacturer’s tested finished products exceeded 100 ppb iAs. A separate cooperating manufacturer’s finished brown-rice products averaged 97.62 ppb iAs (n=21), with half exceeding 100 ppb iAs and one product at 129 ppb iAs. Ingredient-level pre-shipment results included rice-derived ingredients tested at 309 ppb total arsenic (organic brown rice flour), 200 ppb tAs (medium-grain whole rice, soft white wheat flour), 91 ppb iAs (Flour Rice Long Grain), and 67 batches of rice flour over 90 ppb iAs at a fourth cooperating manufacturer. Non-rice ingredients with high arsenic results included a vitamin/mineral pre-mix at 223 ppb iAs.

Inorganic arsenic — additive enzymes. Pre-shipment testing of baking and bread-improver enzymes used by a cooperating manufacturer reported arsenic concentrations of 913.4 ppb (amylase, 9/19/2018), 741.1 ppb (amylase, 4/26/2018), 710.9 ppb (BAN 800, 10/7/2017), 679.0 ppb (alpha amylase, 11/29/2017), 645.1 ppb (amylase, 10/12/2017), and 583.6 ppb (Sebamyl 100, 8/20/2017). The report identifies BAN 800 as an enzyme used to improve crumb softness in baked goods. The acceptance specifications for these additive enzymes were set at ≤3000 ppb arsenic and ≤500 ppb cadmium, well above any food-product regulatory ceiling.

Lead — ingredients. Pre-shipment testing identified spice ingredients as the highest-lead inputs into baby food: cinnamon at 886.9 ppb Pb (10/19/2016), organic cumin at 644.9 ppb Pb (5/21/2018), organic coriander at 603.5 ppb Pb (8/11/2017), oregano at 570.4 ppb Pb (10/11/2016), and additional cinnamon and cumin lots in the 100-230 ppb Pb range. A vitamin/mineral pre-mix tested at 352 ppb Pb. Among non-spice ingredients with elevated lead: enzyme additives (alpha amylase 114.5 ppb Pb, amylase 108.8 ppb Pb), organic lemon (102 ppb Pb), sunflower lecithin (54.9-71.6 ppb Pb), sweet potato (up to 55.3 ppb Pb), quinoa flour and quinoa seeds (21-25 ppb Pb), prune (27.9-41.5 ppb Pb), apricot (28 ppb Pb), mango (32.3 ppb Pb), dehydrated potato (32.4 ppb Pb), blueberry (22.7 ppb Pb), organic pears (21.7 ppb Pb), and carrots (up to 20 ppb Pb).

Lead — finished products. Finished baby-food products tested up to 641 ppb Pb. Of 206 finished products tested by one cooperating manufacturer, 39 (18.9%) tested over 10 ppb Pb and 16 (7.8%) tested over 20 ppb Pb. Across product-form categories in the finished-product Pb table, the spot-check ranges were: pouch and jar purées with apple-spinach-pea-and-kiwi or apple-spinach bases up to 43 ppb Pb; pea-and-spinach teething wafers 18-23 ppb Pb; apple-and-broccoli puff snacks 5.8-11 ppb Pb; strawberry-and-beet puff snacks 7.8-10 ppb Pb; banana-and-pumpkin puff snacks 5.5-13 ppb Pb; multi-grain cereal canisters 5.2 ppb Pb; and one canister-format purple-carrot-and-blueberry purée at the table maximum of 641 ppb Pb. Most simple single-ingredient purées (sweet-potato jars, carrot jars, mango, pear-prune, green-bean) read at the table’s <4.0 ppb Pb floor.

Lead — fruit juice concentrates. Pre-shipment testing of grape juice concentrate and apple juice concentrate reported lead concentrations at 21-48 ppb Pb across multiple lots. The cooperating manufacturer’s average across tested juice concentrates was 11.2 ppb Pb, and over 83% of tested juice concentrates contained more than 1 ppb Pb.

Cadmium — ingredients. Pre-shipment testing identified spice cadmium values up to 344.5 ppb Cd (cinnamon, 10/19/2016), with multiple cinnamon and organic cinnamon lots in the 117-225 ppb Cd range, organic garlic at 186 ppb Cd, oregano at 176.5 ppb Cd, dehydrated potato lots at 119.6-148.4 ppb Cd, spinach (fresh and pureed) at 117-143 ppb Cd, and organic barley flour at 260 ppb Cd. Other cadmium-elevated ingredients included organic chopped broccoli (250 ppb Cd), organic date paste (190-220 ppb Cd), organic cinnamon powder (200 ppb Cd), organic brown flax milled (130-190 ppb Cd), organic yellow papaya puree (170 ppb Cd), organic whole wheat fine flour (160 ppb Cd), organic red lentils (130 ppb Cd), organic oat flakes (130 ppb Cd), organic barley flour (110 ppb Cd), and organic oat flour (102 ppb Cd).

Cadmium — finished products and carrots. One cooperating manufacturer reported 65% of its finished baby-food products at over 5 ppb Cd, with a multi-grain cereal at 49 ppb Cd and 125 additional products over 5 ppb Cd. A separate cooperating manufacturer’s testing of conventional carrots reported up to 87 ppb Cd with 75% of tested carrot lots exceeding 5 ppb Cd. The cited EPA and FDA drinking-water cadmium limits are 5 ppb; the EU infant-formula cadmium limit ranges 5-20 ppb.

Total mercury (tHg). Only one of the four cooperating manufacturers regularly tested for mercury. That manufacturer’s finished baby-food products tested up to 10 ppb tHg, with two additional products at 9.8 and 7.3 ppb tHg, and 56 finished products over 2 ppb tHg in total. Beech-Nut and Hain reportedly did not test for mercury at all. Gerber reportedly tested mercury only on carrots, sweet potatoes, and lemon juice concentrate.

Confidential Hain-FDA presentation, August 1, 2019. Hain submitted a slide presentation to FDA showing that in 100% of finished Hain baby foods tested, inorganic arsenic levels ran 28-93% higher than ingredient testing alone predicted. Hain’s average finished-product iAs was 97.62 ppb. Half of Hain’s brown-rice finished products tested at or above 100 ppb iAs; one product was 129 ppb iAs. Hain attributed the finished-product/raw-material divergence to its vitamin/mineral pre-mix additive: Hain’s testing showed that pre-mix at 223 ppb iAs and 352 ppb Pb was approved on a “deviation” basis using a theoretical-calculation argument that the 2.08% inclusion rate would dilute the additive’s contribution below 100 ppb in the finished good. The report notes that Hain did not confirm finished-product compliance by direct measurement after the deviation approval.

Internal corporate threshold values. Cooperating manufacturers’ internal pre-shipment specifications referenced in the report include: 100 ppb iAs goal threshold for Cereals with <75% rice and 115 ppb iAs goal threshold for Cereals with >75% rice (one cooperating manufacturer’s Jan-2019 specification); 75 ppb iAs as-sold for Infant Formula; 50 ppb Pb as-sold for Infant Formula and 50 ppb Pb as-consumed for Cereals (claimed January-2019 lowering from 100 ppb); 10 ppb Cd as-sold for Infant Formula; 10 ppb tHg as-consumed for Cereals and Wet Foods; 3,000 ppb iAs and 3,000 ppb Cd ingredient specs for additives like vitamin mix at a second cooperating manufacturer; 5,000 ppb Pb ingredient spec for enzyme additives like BAN 800 at the same manufacturer; 200 ppb As/Pb/Cd ingredient specs for some flours at a third cooperating manufacturer. Multiple internal records show approved “deviation” entries above the manufacturer’s own specification limit.

Cited comparator regulatory thresholds. FDA bottled-water limits: 10 ppb iAs, 5 ppb Pb, 5 ppb Cd. EPA drinking-water action level: 15 ppb Pb. EPA drinking-water mercury cap: 2 ppb. FDA juice draft guidance (never finalized): 10 ppb iAs in apple juice. FDA infant-rice-cereal action level: 100 ppb iAs (finalized August 2020). FDA candy guidance: 100 ppb Pb. FDA juice limit: 50 ppb Pb. EU infant-formula lead limit: 20 ppb Pb. EU infant-formula cadmium limit: 5-20 ppb Cd (range). WHO drinking-water guidelines: 10 ppb Pb (provisional), 3 ppb Cd. AAP school-fountain lead recommendation: 1 ppb Pb. Consumer Reports fruit-juice lead: 1 ppb Pb. Environmental Defense Fund baby-food lead: 1 ppb Pb. Healthy Babies Bright Futures: “no measurable amount” goals for iAs, Pb, Cd, tHg in baby food.

Manufacturer-level cooperation status. Four manufacturers cooperated and produced records: Nurture, Inc. (HappyBABY); Beech-Nut Nutrition Company; Hain Celestial Group (Earth’s Best Organic); and Gerber Products Company. Three did not cooperate: Walmart Inc. (Parent’s Choice), Sprout Foods Inc. (Sprout Organic Foods), and Campbell Soup Company (Plum Organics). For the non-cooperating manufacturers, the report cites Healthy Babies Bright Futures 2019 independent retail-purchased testing of representative products: across the five reproduced HBBF rows covering multigrain teething wafers, multigrain soft-baked snack bars, strawberry rice rusks, quinoa puff snacks, and strawberry-yogurt cereal snacks, finished-product concentrations spanned 40-108 ppb tAs, up to 66 ppb iAs (where iAs was speciated; two of five rows reported only tAs), 1.4-39.3 ppb Pb, 2.4-41.5 ppb Cd, and <0.137-2.05 ppb tHg. The highest-contamination row across these five was a rice-and-quinoa puff snack; the highest single iAs value was on a rice-rusk product. Independent retail testing of the non-cooperating manufacturers thus documented finished-product values comparable to or higher than the cooperators’ internal records for the same product forms.

Methods (brief)

The data underlying the report are internal company laboratory records subpoenaed by the Subcommittee under its oversight authority and produced between November 2019 and December 2019 by the four cooperating manufacturers. Analytical methods are not standardized across manufacturers; each company’s contract laboratories and internal QC procedures generated the values, with units consistently reported in parts per billion (ppb, equivalent to µg/kg). Speciation of arsenic is variable: some pre-shipment specifications and results report inorganic arsenic explicitly (with specifications in the 100-115 ppb range), while other specifications and results report total arsenic only (with specifications up to 3000 ppb for enzymes). The report does not provide LODs, certified reference material results, or interlaboratory validation for the company-produced datasets. Finished-product results are reported “as sold” or “as consumed” depending on the product category and the manufacturer’s internal convention.

For the Hain-FDA presentation (August 1, 2019), Hain ran paired finished-product and raw-material analyses on rice-based products under FDA observation; the presentation is identified as Hain’s submission “FDA Testing Result Investigation” (file Hain-000154 in the production). The presentation’s headline finding — that finished-product iAs runs 28-93% higher than ingredient testing alone predicts — is the central data integrity claim in the report.

Limitations to flag for downstream use: (1) the data are company-internal and not independently verified by an accredited reference lab; (2) speciation is inconsistent across rows, so iAs/tAs distinctions must be tracked at the row level not at the report level; (3) the report uses “<9.54 ppb” and similar non-detect placeholders without disclosing LOD/LOQ values; (4) sample identifiers are de-anonymized at the product-name level for finished products but reduced to commodity-name level (e.g., “Org Brown Rice Flour”) for ingredients, with lot/preshipment numbers redacted in some excerpted tables.

Implications

Certification: Provides finished-product-vs-raw-material divergence evidence (28-93% higher iAs in finished baby foods than in the raw ingredients used to make them, per the Hain-to-FDA presentation) relevant to standards set on a raw-material basis. Contributes occurrence data for finished baby cereal, baby snack, fruit purée, root and non-root vegetable purée, and fruit-juice categories, and ingredient-level occurrence data for rice, rice flour, brown-rice flour, spices, vitamin/mineral pre-mix additives, and food-enzyme additives.

Courses: The report is the canonical case study for explaining why ingredient testing does not substitute for finished-product testing, and why corporate internal thresholds that exceed every analogous regulatory standard (drinking water, candy, juice, infant formula) are a market-failure signal. Course content on supply-chain quality assurance, brand QA program design, and regulatory affairs strategy for infant-food categories should include this report.

App: The ingredient-level findings (rice, brown rice flour, cinnamon, cumin, oregano, vitamin/mineral pre-mix, sweet potato, carrots) are direct inputs to the consumer app’s ingredient-driven contamination-likelihood model. The finished-product results are useful for calibrating the residual-uncertainty term in the app’s confidence intervals.

Verification notes

Fresh ingest 2026-06-02 from raw/manual-fetch/Kimi_Agent_Download Corruption Issue/condiments2_papers/05_Snacks_Canned_Prepared/House_Baby_Food_Heavy_Metals_Report_2021.pdf. PDF SHA256 0c7404630cf3a7a92b2e89c6bd20fe4930c9f54406facfc087631ba07ebad0d7. Read in full (59 pages including TOC, exec summary, four metal sections, four manufacturer sections, three non-cooperating-manufacturer sections, FDA-failure section, recommendations, conclusion). All ingredient-level pre-shipment test values quoted above were cross-checked against the report’s excerpted tables at pages 17-18 (Beech-Nut arsenic), 24-26 (Beech-Nut lead), 30-31 (Beech-Nut cadmium and Hain cadmium), 16 (Hain arsenic), 19-21 (Gerber rice flour arsenic), 27-28 (Gerber juice concentrate lead), 41 (Hain vitamin pre-mix deviation), 14-15 (Nurture iAs finished products), 22-23 (Nurture Pb finished products), 31-32 (Nurture Cd and Hg finished products), 43-47 (Walmart/Campbell/Sprout HBBF data). Brand-firewall handling per Part 12 Exception 1: companies are named as subjects of the Congressional investigation (a public-record regulatory event), but per-manufacturer ranking tables are not reproduced and the page’s primary scientific content is presented at the ingredient and analyte level. The report’s narrative explicitly identifies four cooperating and three non-cooperating manufacturers; that cooperation status is a regulatory-event fact, not a contamination ranking, and is preserved verbatim.

Audit subagent (2026-06-02) returned QUARANTINE with three ❌ findings; three findings were verified and applied as follows. (1) Finding “Blueberry Beet Rice Cakes at 14 ppb Pb is not in the source”: verified correct against PDF page 36, which shows Blueberry Beet Rice Cakes at <4.0 ppb Pb on 10/14/19; the 14 ppb value was a misread, removed. (2) Finding “Stage 2 Apple Spinach Pea and Kiwi descriptor is invented; PDF uses Baby 7+ Months”: verified correct, the Stage 2 qualifier was removed, and the branded SKU was replaced with a product-form descriptor. (3) Finding “Part 12 brand firewall violated by branded-SKU + value enumeration (both HappyBABY/Nurture finished-product spot-checks and HBBF non-cooperator product values)”: verified correct under the strict 2026-05-17 reading; regulatory-event Exception 1 covers naming the manufacturers as investigation subjects, not reproducing branded-product-name + value rows. Both enumeration paragraphs were rewritten to use product-form aggregations (apple-and-broccoli puff snacks, strawberry-and-beet puff snacks, pea-and-spinach teething wafers, multigrain teething wafers, rice-rusk snacks, quinoa-and-rice puff snacks, multigrain soft-baked snack bars, strawberry-yogurt cereal snacks) while preserving the cooperation-status fact. Finding “ingredients/cumin is not in the taxonomy snapshot” was rejected as a false positive: wiki/ingredients/cumin.md exists as a provisional scaffold (created 2026-06-02 by ingest of sabri2025-moroccan-spices-casablanca-metals), so the slug is valid; the audit subagent’s docs/gpt-collaboration/taxonomy-snapshot.md reference is stale relative to today’s autonomy-stub additions. Check 5 ⚠️ on the Certification implication paragraph (methodology prescription tone) was accepted and the paragraph was trimmed to occurrence-data framing.

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.