Heavy metals in Infant Formula, Powder (Non-Soy)

Overview

This page is the public evidence page for powdered non-soy infant formula. It is organized around source-backed product concentration evidence, with exposure estimates and broad formula studies separated so they do not get mistaken for direct ppb product measurements.

Row Relationship

This row is the clean-benchmark counterpart to Infant Formula, Powder (Soy-Based) for the row architecture relationship covering aluminum (Al), nickel (Ni), and cadmium (Cd).

Evidence summary

The table below summarizes what the peer-reviewed and government literature cited on this page reports for heavy-metal concentrations in non-soy (cow milk-based), powder infant formula. Values are pulled directly from cited sources without re-aggregation. This page publishes literature evidence only, not certification thresholds.

Methodology rules for speciation, basis preservation, non-detect handling, and source pooling are stated in the Methodology section above and apply to every row below.

Source Evidence Inventory

This table lists what each source actually reports. Highest values are source-scope observations, not public certification limits or cross-source standards.

Citations in this table use short numeric labels. Each number is a clickable link to that source’s page, and the full citation list with titles, years, and source types is in the Sources section at the bottom of this page. Quick legend for the numbers used here and in the prose sections below: 1 Chung 2021 (China cow-milk formulas), 2 Almeida 2022 (Brazil cow-milk formulas), 3 FSA 2016 (UK formula and infant foods survey), 4 Dabeka 1987 (Canada milk-base historical), 5 Pandelova 2012 (EU pooled market baskets), 6 Jackson 2012 (formulas without brown rice syrup), 7 Gardener 2019 (U.S. exposure estimates), 8 Collado-López 2025 (global scoping review), 9 Signes-Pastor 2018 (infant arsenic biomarkers), 10 Meli 2024 (Italy baby-food analytical), 13 Soares 2000 (Portugal hexavalent chromium (Cr-VI) powdered milk).

Measured concentrations

The FDA 2026 special survey is the first source in this row that gives a reconstructable product-label subset for several metals. These values are expressed as prepared for feeding, so they should not be silently pooled with dry-powder-as-sold ppb values. The extraction below is included as a traceability appendix, not as a public standard. Analytical Results for Toxic Elements in Infant Formula, FY2023-FY2025 Special Survey

The Digest formula papers add useful source-scope rows, but they mostly report means, medians, ranges, or maxima rather than full product-level distributions. These rows support the evidence pool and show high-end source context; they do not by themselves set public limits.

Supporting evidence and scope

The sources below are catalogued as product-context candidates for this row. The “Author-scope row-fit” column states what the authors actually resolved on each axis: matrix (cow milk-based, soy-based, rice-based, non-rice, or unresolved) and format (powder, ready-to-feed liquid, concentrated liquid, dry, or unresolved). A source counts toward this row’s evidence pool only once; rows marked “Cross-reference” already appear as direct evidence elsewhere on this page and are not counted again here.

Evidence Used For This Row

The direct row-fit evidence is strongest when a study measures cow-milk, milk-base, or powdered milk formula as a product concentration. Chung 2021, Almeida 2022, FSA 2016, Dabeka 1987, Pandelova 2012, and Soares 2000 are therefore more useful for this row than broad baby-food papers, but each still has limitations that require AI adjudication before standards use.

The largest interpretive issue is scope. China-market, Brazil-market, UK-market, Canada-historical, EU-pooled-basket, and U.S. prepared-for-feeding values should not be silently merged without basis matching, AI row-fit adjudication, and jurisdiction metadata. They can show what has been observed in formula, and they may contribute to an aggregate evidence pool when the standards workflow can document comparability and 95% confidence.

Limits applicable to this category

The category inherits the maximum levels and action levels that apply to it. These are regulatory limits, not certification thresholds.

Federal / Regulatory Limits vs Field Findings

This is the fast comparison view for standards developers, regulators, retailers, brands, and legal teams. It shows the applicable federal or regulatory limit next to the current field-evidence state. It is not an HMTc pass/fail table; technical distributions remain in the evidence sections below.

Lead Benchmark Context

HMI normalizes this row’s lead benchmarks to ppb so regulatory ceilings, exposure screens, and occurrence values can be compared on one concentration scale. The values below do not all mean the same thing: FDA and EU entries are regulatory context, Prop 65 is a serving-based exposure screen, and source tables on this page remain occurrence evidence.

No U.S. FDA formula-specific lead action level is currently established; the EU powder ceiling is a legal backstop, not a clean-product target.

Full crosswalk: Category 1 Lead Benchmark Context.

Ingredient drivers

The inputs below are ordered by their share of the routed occurrence evidence for the metal each drives; the bar shows relative position within this category, not a comparison against any limit.

Sources of variation

The promoted formula scoping review separates cow-based, soy-based, specialty, and nonspecified formulas, but it does not resolve powder-versus-ready-to-feed differences for this row. 8

The promoted powdered-milk analytical study resolves powder format but does not resolve soy versus non-soy formula. 10

Pandelova 2012 reported higher cadmium values in pooled soy-formula baskets than in pooled milk-formula and hypoallergenic-formula baskets, but pooled market baskets cannot be used as an individual-product percentile distribution. 5

Potential variance drivers for non-soy powdered formula should be documented only after sources distinguish formulation, ingredient inputs, processing equipment, packaging, and analytical method.

Controls and mitigation

Infant formula is a manufactured product whose heavy-metal burden is determined by its ingredient inputs (dairy or soy protein base, vitamin and mineral premix, processing water, and processing equipment contact surfaces) rather than by whole-food agricultural contamination alone. The contamination profile of the finished product reflects the aggregate of all these inputs. Levers are ordered by approximate impact magnitude based on what the cited literature supports.

Agronomic levers: not applicable to this product category as a direct lever. Agronomic interventions on dairy herds live upstream on the relevant ingredient pages.

Cross-links: Infant Formula Powder if it exists; relevant mitigation pages where they exist.

Testing and verification posture

Occurrence values in this category are measured by ICP-MS in the source literature, the same method the FDA Toxic Elements Program uses for the survey data behind its action levels.

Exposure Estimates From Formula Consumption

Gardener 2019 is useful because it reports infant exposure estimates from formula consumption. It is not a product concentration table and should not be used as ppb product-limit evidence.

French TDS Category Rows

Chekri 2019 reports French infant formula, follow-on formula, and growing-up milk values as consumed after preparation; it does not separate powder from ready-to-feed, soy from non-soy, or cow-milk from other formulas, so these rows support broad formula occurrence context rather than a row-specific product distribution. Chekri 2019

Why This Category Is High-Risk

A 2025 global scoping review of baby foods and infant formulas reported heavy-metal detections in 63 percent of evaluated infant-formula determinations, with lead (Pb), Cd, arsenic (As), and mercury (Hg) each detected in formula items; in the review’s primary-protein-source subgrouping, Pb was detected in 73 percent of cow-based formula items and Cd in 44 percent of cow-based formula items. 8

A 2012 arsenic speciation study reported total arsenic concentrations of 2 to 12 ng/g in 15 infant formulas without organic brown rice syrup; because 1 ng/g equals 1 ug/kg, this corresponds to 2 to 12 ppb total arsenic in formula powder, though the study does not isolate non-soy formula powder as a row-specific category. 6

A 2018 infant biomarker study cited prior work reporting total arsenic in formula powder up to 12.6 ug/kg, but the study does not separate soy-based from non-soy powdered formula. 9

A 2024 analytical study of European baby foods included powdered milk and reported that cadmium and lead were below the study LOD in all samples, while mercury was detectable in all samples and one powdered-milk sample had the highest estimated nickel intake in the study at 9.43 ug/kg body weight per day. 10

Non-soy-specific risk characterization remains incomplete.

How The App Would Estimate Risk From An Ingredient List

The app model placeholder for this row should treat Infant Formula Powder and Infant Formula, Powder (Non-Soy) as unresolved ingredient targets until source-backed contamination profiles exist.

How standards math uses this page

This page documents what the cited sources report. The row-standard is an aggregate across all contributing sources after basis adjustment and row-fit review — not a decoration on any individual source row — and is set separately under the certification program, not published on this public page.

Citing this page at a single source’s maximum value as if it were a threshold justification misreads the evidence architecture: the maximum observed in one study is not the same as a representative value across the full source pool. HMT&C certification threshold decisions are made separately under the certification program and are not published on this public page.

Historical Recalls/Enforcement

See the page-level crosswalk above and Regulatory Crosswalk vs Field Findings for current regulatory context; row-specific enforcement events remain pending.

No row-specific regulatory event has been added for this scaffold.

Historical recalls and enforcement

FDA Closer to Zero infant-and-young-child food enforcement actions are the dominant Cat 1 regulatory-event context: the 2023 WanaBana cinnamon-applesauce Pb-chromate adulteration outbreak (detailed in Herbal botanicals and the Napier 2024 MMWR / Troeschel 2024 reports) prompted FDA Import Alert 99-42 (FDA 2024). Other Cat 1 regulatory events of note: the longstanding HBBF “Baby Food Heavy Metals” reports (Houlihan 2019) and 2021 US House Subcommittee report drove FDA’s Closer to Zero action-level rulemaking (FDA 2025, FDA 2020). Individual brand recall actions are not enumerated here; the recalls are framed as regulatory events that established the action-level framework currently in effect.

Contradiction watch

The living-review detector has flagged 24 contributing source value(s) that disagree with the current synthesis by more than 2× the tolerance band. A re-synthesis pass for the affected (ingredient, metal) cell(s) is warranted; the synthesis claim is not retracted by this flag.

Full per-flag audit at data/evidence/synthesis-contradictions.csv. Trigger is documented in.

Measurements filed from the retired ingredient page

These sources were cited on the ingredient record before that record was retired.

• Toxicological Profile for Aluminum
• Toxicological Profile for Aluminum
• Mechanism of Iron Ion Homeostasis in Intestinal Immunity and Gut Microbiota Remodeling
• Effect of long-term and short-term imbalanced zinc manipulation on gut microbiota and screening for microbial markers sensitive to zinc status
• The gut microbiome is required for full protection against acute arsenic toxicity in mouse models
• Scientific Opinion on Arsenic in Food
• Human supplementation with Pediococcus acidilactici GR-1 decreases heavy metal levels through modifying the gut microbiota and metabolome
• Sex-specific associations of infants' gut microbiome with arsenic exposure
• Public Health and Paediatric Risk Assessment of Aluminium, Arsenic and Mercury in Infant Formulas Marketed in Nigeria
• Toxic trace elements in different brands of milk infant formulae in Nairobi market, Kenya
• Safety assessment of essential and toxic metals in infant formulas
• Metal availability shapes early life microbial ecology and community succession

Evidence gaps

The strongest row-fit evidence on this page is cow-milk formula concentration data from China, Brazil, the United Kingdom, Canada, and pooled European market baskets, plus a 2026 FDA prepared-for-feeding sample-level dataset of 230 cow-milk powder formulas covering lead, cadmium, total arsenic, and total mercury. The peer-reviewed literature and government datasets cited below give consistent direct evidence for those four analytes on this matrix.

Remaining gaps in the cited literature are concentrated in three places. First, nickel, aluminum, hexavalent chromium, tin, and inorganic arsenic in non-soy powdered formula are documented mostly by summary statistics (means, medians, or ranges) rather than by sample-level distributions, which limits how confidently the literature can characterize the upper end of each analyte. Second, methylmercury in formula is a documented data gap; total mercury values cannot substitute because the toxicology of methylmercury and inorganic mercury differs. Third, several cited sources do not separate powder from ready-to-feed liquid, do not separate non-soy from soy-based formula, or report exposure estimates rather than product concentrations; those sources can support context but cannot anchor a non-soy powder distribution on their own.

Last updated 2026-05-08. Sixteen sources cited, all A-tier (peer-reviewed studies or government reports).

Methodology

This page reports what the cited sources say about heavy-metal concentrations in non-soy powdered infant formula. The summary tables and inventories below are governed by a fixed set of methodology rules so the evidence is interpretable and auditable. The rules are stated here once and referenced from every section that depends on them.

Speciation is treated as non-substitutable. Inorganic arsenic (iAs) and total arsenic (tAs) are reported separately, because the toxicology and the regulatory ceilings for each are different. Methylmercury (MeHg) and total mercury (tHg) are reported separately for the same reason; tHg is not used as a proxy for MeHg, and a tHg measurement does not close a MeHg evidence gap. Total chromium (Cr) is not interpreted as hexavalent chromium (Cr-VI) unless the source explicitly speciates Cr-VI.

Basis is preserved and labeled, never silently converted. Concentrations in formula can be reported on at least three bases: powder as placed on the market, powder prepared for feeding (reconstituted with water at the manufacturer-specified dilution), or formula as consumed by an infant. Values on different bases are not directly comparable. Each table below labels the source basis explicitly. Where a basis conversion is needed for context, the conversion factor and assumptions are stated alongside the converted value, and the converted value is marked as indicative rather than direct.

Non-detect handling. Where a source reports a value below its limit of detection (LOD) or limit of quantification (LOQ), this page preserves the source’s reported handling. The 2026 FDA special survey is summarized below using the lower-bound convention <LOD = 0, which is conservative for upper-tail statistics. Other sources are summarized as the source itself reported (mean, median, range, or category statistics).

Source pooling is avoided. Aggregate statistics are not computed by pooling sample values across sources whose LOQs, sampling periods, geographies, and analytical bases differ. When the page reports source-level summary statistics, those summaries are the source’s own, not a re-aggregation across the cited literature. Cross-source pooling, when needed for standards work, is performed in staff tooling with documented basis adjustments and is not published on this page.

Row-fit and matrix-axis labeling. Sources are classified by how cleanly their reported scope matches non-soy powdered infant formula. Matrix-axis labels are exact when the source explicitly identifies the protein basis (cow milk based, milk based, or non-soy) and partial or unknown when the source uses a broad term (“infant formula”) or does not split soy from non-soy. Format-axis labels are exact when the source identifies powder, partial when the source mixes powder with ready-to-feed liquid, and unknown when the source does not state. Row-fit determines whether a source contributes direct evidence or supporting context.

Evidence tiers. A-tier sources are peer-reviewed primary studies, government reports, and authoritative meta-analyses. B-tier sources are industry white papers, NGO reports, and trade publications. C-tier sources are news, blog, and press material. Synthesis on this page leans on A-tier. Every cited source on this page is A-tier as of the last update.

Source coverage. Low denotes 1–2 contributing sources, medium 3–10, and high more than 10. These counts describe coverage, not study quality, agreement, or scientific confidence.

HMT&C threshold-setting is separate. HMT&C certification thresholds for products in this row are developed under the certification program at heavymetalcertified.com, not on this page. The values on this page are literature evidence, not certification thresholds.

Evidence Governance

Public evidence label: Modeled or limited evidence.

This page is part of the Category 1 Evidence Fitness pilot. It summarizes source-backed occurrence evidence, partial distributions, and data gaps for this product row. Existing cited tables remain public page-level synthesis; value-level extraction and standards review are tracked separately in staff tooling.

This page does not publish or justify HMT&C certification limits. Public Index pages show what the cited sources say, what is still uncertain, and where readers can verify the evidence trail.

References

Works cited in this page’s text, in first-appearance order. See Sources for this page’s source inventory. Each title links to a source record with bibliographic details, reported findings, and document provenance.

1. Analytical Results for Toxic Elements in Infant Formula, FY2023-FY2025 Special SurveyU.S. Food and Drug Administration · FDA analytical results table · 2026 · www.fda.govDataset
2. Trace element contents in foods from the first French Total Diet Study on infants and toddlersRachida Chekri, Emilie Le Calvez, Julie Zinck, Jean-Charles Leblanc, Veronique Sirot, Marion Hulin, et al. · Journal of Food Composition and Analysis · 2019 · doi.org/10.1016/j.jfca.2019.02.002Peer-reviewed
3. Concentrations of Heavy Metals in Processed Baby Foods and Infant Formulas Worldwide: A Scoping ReviewSonia Collado-Lopez, Maria Fernanda Rodriguez Hernandez, Rosa Maria Mariscal-Moreno, Martha Maria Tellez-Rojo, Larissa Betanzos-Robledo, Moises Reyes Luna, et al. · Nutrition Reviews · 2025 · doi.org/10.1093/nutrit/nuaf138Peer-reviewed
4. Arsenic, Organic Foods, and Brown Rice SyrupJackson BP, Taylor VF, Karagas MR, Punshon T, and Cottingham KL · Environmental Health Perspectives · 2012 · doi.org/10.1289/ehp.1104619Peer-reviewed
5. Infants’ dietary arsenic exposure during transition to solid foodSignes-Pastor AJ, Cottingham KL, Carey M, Sayarath V, Palys T, Meharg AA, et al. · Scientific Reports 8(1):7114 · 2018 · doi.org/10.1038/s41598-018-25372-1Peer-reviewed
6. Chemical characterization of baby food consumed in ItalyMeli MA, Desideri D, Sisti D, Fagiolino I, and Roselli C · PLOS ONE · 2024 · doi.org/10.1371/journal.pone.0297158Peer-reviewed
7. Ca, Cd, Cu, Fe, Hg, Mn, Ni, Pb, Se, and Zn contents in baby foods from the EU market: Comparison of assessed infant intakes with the present safety limits for minerals and trace elementsPandelova M, Levy Lopez W, Michalke B, and Schramm KW · Journal of Food Composition and Analysis · 2012 · doi.org/10.1016/j.jfca.2012.04.011Peer-reviewed
8. Childhood Lead Exposure Linked to Apple Cinnamon Fruit Puree Pouches — North Carolina, June 2023–January 2024Napier MD and et al. · MMWR Morbidity and Mortality Weekly Report · 2024 · doi.org/10.15585/mmwr.mm7328a2Government
9. Investigation of Lead and Chromium Exposure After Consumption of Contaminated Cinnamon-Containing Applesauce — United States, November 2023–April 2024Troeschel AN, Buser MC, Winquist A, Ruckart P, Yeh M, Kuai D, et al. · Morbidity and Mortality Weekly Report (MMWR) · 2024 · doi.org/10.15585/mmwr.mm7414a2Government
10. FDA Import Alert 99-42: Detention Without Physical Examination of Spices Due to Lead ContaminationU.S. Food and Drug Administration · FDA Import Alerts · 2024 · www.accessdata.fda.govRegulation
11. What’s in My Baby’s Food? A National Investigation Finds 95 Percent of Baby Foods Tested Contain Toxic Chemicals That Lower Babies’ IQ, Including Arsenic and LeadJane Houlihan and Charlotte Brody · Healthy Babies Bright Futures · 2019 · www.hbbf.orgNGO
12. Action Levels for Lead in Processed Food Intended for Babies and Young Children: Guidance for IndustryU.S. Food and Drug Administration · U.S. Department of Health and Human Services, Food and Drug Administration, Human Foods Program · 2025 · www.fda.govGuidance
13. Inorganic Arsenic in Rice Cereals for Infants: Action Level, Guidance for IndustryU.S. Food and Drug Administration · 2020 · www.fda.govGuidance
14. Toxicological Profile for AluminumAgency for Toxic Substances and Disease Registry · 2008 · www.atsdr.cdc.govGovernment
15. Mechanism of Iron Ion Homeostasis in Intestinal Immunity and Gut Microbiota RemodelingBao H, Wang Y, Xiong H, Xia Y, Cui Z, and Liu L · International Journal of Molecular Sciences · 2024 · doi.org/10.3390/ijms25020727Peer-reviewed
16. Effect of long-term and short-term imbalanced zinc manipulation on gut microbiota and screening for microbial markers sensitive to zinc statusChen L, Wang Z, Wang P, Yu X, Ding H, Wang Z, et al. · Microbiology Spectrum · 2021 · doi.org/10.1128/Spectrum.00483-21Peer-reviewed
17. The gut microbiome is required for full protection against acute arsenic toxicity in mouse modelsCoryell M, McAlpine M, Pinkham NV, McDermott TR, and Walk ST · Nature Communications · 2018 · doi.org/10.1038/s41467-018-07803-9Peer-reviewed
18. Scientific Opinion on Arsenic in FoodEFSA Panel on Contaminants in the Food Chain (CONTAM) · EFSA Journal 2009;7(10):1351 · 2009 · doi.org/10.2903/j.efsa.2009.1351Government
19. Human supplementation with Pediococcus acidilactici GR-1 decreases heavy metal levels through modifying the gut microbiota and metabolomeFeng P, Yang J, Zhao S, Ling Z, Han R, Wu Y, et al. · npj Biofilms and Microbiomes · 2022 · doi.org/10.1038/s41522-022-00326-8Peer-reviewed
20. Sex-specific associations of infants’ gut microbiome with arsenic exposureAnne G. Hoen, Juliette C. Madan, Zhigang Li, Modupe Coker, Sara N. Lundgren, Hilary G. Morrison, et al. · Scientific Reports · 2018 · doi.org/10.1038/s41598-018-30581-9Peer-reviewed
21. Public Health and Paediatric Risk Assessment of Aluminium, Arsenic and Mercury in Infant Formulas Marketed in NigeriaIgweze ZN, Ekhator OC, Nwaogazie I, and Orisakwe OE · Sultan Qaboos University Medical Journal 20(1):e63-e70 · 2020 · doi.org/10.18295/squmj.2020.20.01.009Peer-reviewed
22. Toxic trace elements in different brands of milk infant formulae in Nairobi market, KenyaV. O. Odhiambo, R. Wanjau, J. O. Odundo, and M. P. Nawiri · African Journal of Food Science · 2015 · doi.org/10.5897/AJFS2015.1315Peer-reviewed
23. Safety assessment of essential and toxic metals in infant formulasSipahi H, Eken A, Aydın A, Şahin G, and Baydar T · The Turkish Journal of Pediatrics 56(4):385-391 · 2014 · www.turkishjournalpediatrics.orgPeer-reviewed
24. Metal availability shapes early life microbial ecology and community successionJoshua Soto Ocaña, Elliot S. Friedman, Orlaith Keenan, Nile U. Bayard, Eileen Ford, Ceylan Tanes, et al. · mBio 15(7):e00854-24 · 2024 · doi.org/10.1128/mbio.01534-24Peer-reviewed

Sources

Source records associated with this topic. Inclusion does not establish that every finding applies to this product or ingredient. References above identify works cited in the text. Additional records await source or bibliographic review and are excluded from this list.

Update history

No substantive edit history is available in this build. The full commit record is available in git.