Summary
PROSPERO-registered systematic review examining contaminant concentrations in human milk (HM) and infant formula (IF) in the United States, commissioned by the FDA’s Human Foods Program. The review is companion to O’Connor et al. 2026 (exposure and bioavailability series). This paper covers contaminant concentrations in HM and IF matrices directly, while O’Connor et al. 2026 covers what those concentrations mean for infant biomarker exposure. For the ready-to-feed formula page, this review is the source for understanding the state of evidence on arsenic, cadmium, lead, mercury, and PFAS in US infant formula.
Key numbers
Human milk (US, N=14 articles from 15 studies):
Contaminant concentrations in HM are summarized in Table 2 of the paper:
| Contaminant | Collection timing | Concentration (µg/L) | n |
|---|---|---|---|
| Total arsenic | 2-7 wk postpartum (Carignan 2015) | mean 0.33 ± 0.17 | 9 |
| Total arsenic | 2 wk-2 y postpartum (Klein 2017) | mean 3.47 ± 0.84; NR/20 detected | 20 |
| Cadmium | 2 wk-2 y (Klein 2017) | not detected, 0/20 | 20 |
| Cadmium | 1968 (Murthy 1971) | 19 ± 27; NR/22 | 22 |
| Lead | 1.5 mo (Sowers 2002) | 6.1 ± 1.1 (SE) | 15 |
| Lead | 1-6 mo combined (Rabinowitz 1985) | 17 ± 2 (SE); NR/100 | 100 |
| Mercury | 56+ d (Pitkin 1976) | 0.93 ± 0.23 (SE); 14/32 | 32 |
Key US-specific finding: “The evidence did not support a conclusion about the composition and variability of [arsenic, cadmium, lead] in [HM] in the United States. This was due to small sample sizes from 1 geographical region and little detail on demographic characteristics to determine generalizability to the entire US population.”
Infant formula (US and high-HDI, N=16 articles):
For IF, the review identified “small and unrepresentative samples” and noted most evidence was published before 2000. PFOA and PFOS were detected in US HM samples. PFAS were largely undetected in IF. No contemporary (post-2000) nationally representative US IF concentration data for arsenic, cadmium, lead, or mercury were identified.
GRADE certainty: Moderate for PFOA and PFOS detected in HM. Low or insufficient for arsenic, cadmium, lead, mercury in both HM and IF.
Methods (brief)
PROSPERO-registered (CRD42024528756). Databases: CAB Abstracts, CENTRAL, CINAHL, Embase, MEDLINE through 2 April 2025. Eligible studies: published literature assessed contaminant concentrations in HM or IF from US-based samples or from countries rated “high” or “very high” HDI. Two independent reviewers screened records; ROB 2 for RCTs, ROBINS-E for observational. 332 total articles included; 15 articles from 14 US-based studies selected for the US-specific analysis (HM); 16 IF articles selected. GRADE certainty of evidence assessed by TEP. Narrative synthesis only; meta-analysis not conducted due to heterogeneity.
Limitations
Most studies are from a single geographic region (New England, particularly New Hampshire). Most lead studies were conducted before 2000, limiting relevance to contemporary formula formulations. No studies reported speciated arsenic (inorganic vs organic) in IF. Mercury speciation (methylmercury vs total mercury) was not assessed in IF studies. No contemporary nationally representative US data on arsenic, cadmium, lead, or mercury in IF were identified.
Implications
- Certification: This SR documents a gap in contemporary, nationally representative US data on heavy metals in infant formula. The absence of an identified IF concentration dataset for arsenic, cadmium, lead, and mercury in the United States is itself a defensible justification for HMTc using international occurrence data to set thresholds.
- Courses: Useful as a teaching example of how the absence of a national infant formula monitoring program in the US creates evidence gaps that neither the FDA nor independent researchers have filled.
- App: Does not add product concentration data for contemporary US IF.
- Microbiome: No direct microbiome endpoint.
Provenance notes
Open-access CC BY article. PDF retrieved from PMC (PMC12674019). SHA256 verified against MANIFEST.csv for HMI-PROD-ready-to-feed-formula-1.
Update history
No substantive edit history is available in this build. The full commit record is available in git.