Gardener et al. 2019 — Lead and cadmium in 564 US baby foods and infant formulas (Clean Label Project)
This Clean Label Project-funded study is one of the largest single-point examinations of heavy metal contamination in US commercial baby food to date, testing 564 products for Pb and Cd by ICP-MS across eight product sub-categories: infant formula (n=91), cereals (n=30), kids’ meals (n=23), toddler formula (n=22), juices/drinks (n=30), jars/first meals (n=107), pouches (n=140), snacks (n=107), and electrolyte solutions (n=14). Lead was detected in 37% of samples and Cd in 57%. The study’s central finding for infant formula is that none of the 91 formulas exceeded the FDA’s 6 µg/day provisional limit for lead at a 4-month-old’s recommended daily intake (31 oz), but 22% exceeded California Proposition 65’s stricter 0.5 µg/day lead limit, and 23% exceeded Proposition 65’s 4.1 µg/day cadmium limit. Rice-containing products had the highest Pb and Cd concentrations across all baby food sub-categories, and organic certification was not protective.
Key numbers
Full-sample distribution across all 564 products (Table 1):
| Quantile | Cd (µg/kg) | Pb (µg/kg) |
|---|---|---|
| Maximum | 103.90 | 183.60 |
| 99% | 42.50 | 62.75 |
| 95% | 29.44 | 18.50 |
| 90% | 20.75 | 10.80 |
| 75% (Q3) | 9.54 | 5.60 |
| 50% (median) | 2.76 | 0.00 (non-detect) |
| 25% (Q1) | 0.00 (ND) | 0.00 (ND) |
LOQ: 2 µg/kg for Cd, 4 µg/kg for Pb (method-specific). Non-detects imputed as 0 µg/kg in distribution analysis (not ½LOD convention — important for interpreting the medians above).
75th percentile by product sub-category (Fig. 1, solid baby foods n=471):
- Cereals and snacks and kids’ meals had the highest 75th percentile values for both metals
- Cadmium and lead levels varied significantly across food type (Kruskal-Wallis p < 0.0001 for both)
Ingredient associations (top 5 ingredients in product label, n=471 solid baby foods):
- Cadmium highest in products containing rice, quinoa, wheat, and oats (Wilcoxon p < 0.05 for each)
- Cadmium lowest in products containing apples, pears, peaches, and milk
- Lead elevated in products containing rice, quinoa, and sweet potatoes (Wilcoxon p stated as >0.05, a typographical error likely — context implies significance given the narrative; interpret with caution)
- Organic certification: no significant association with either metal (Cd p=0.65; Pb p=0.28)
Infant formula results (n=91, estimated daily exposures at 31 oz/day for 4-month-old, Table 4):
| Quantile | Est. Pb exposure (µg/day) | Est. Cd exposure (µg/day) |
|---|---|---|
| Maximum | 2.68 | 23.33 |
| 95% | 1.06 | 8.04 |
| 90% | 0.78 | 6.11 |
| 75% | 0.43 | 3.86 |
| 50% | 0.00 (ND) | 0.00 (ND) |
Regulatory comparison for infant formula at 31 oz/day:
- 0/91 exceeded FDA 6 µg/day Pb limit
- 22% (20/91) exceeded Prop 65 0.5 µg/day Pb limit
- 14% exceeded WHO PTMI (5.3 µg/day Cd for a 6.4 kg 4-month-old)
- 23% exceeded Prop 65 4.1 µg/day Cd limit
Solid baby food (n=471) — percentage exceeding lead limits:
- FDA 6 µg/day limit: 0.21% in 1 serving; 1.27% in 2 servings; 3.18% in 300 cal; 6.58% in 500 cal
- Prop 65 0.5 µg/day limit: 18% in 1 serving; 26% in 2 servings; 34% in 300 cal; 40% in 500 cal
Solid baby food (n=471) — percentage exceeding cadmium limits:
- WHO 7.5 µg/day (9 kg baby): 0.42% in 1 serving; 1.49% in 2 servings; 2.12% in 300 cal; 6.79% in 500 cal
- Prop 65 4.1 µg/day: 1.49% in 1 serving; 6.16% in 2 servings; 8.07% in 300 cal; 21.44% in 500 cal
Rice prevalence in the sample: Rice appeared in 10% of all baby food samples tested — including 50% of snacks, 47% of cereals, 1% of infant formulas, and 45% of toddler formulas. The authors highlight rice as the ingredient most associated with both lead and cadmium elevation across the product set.
Maximum single-sample values: Pb maximum 183.6 µg/kg; Cd maximum 103.9 µg/kg. Only one sample exceeded 100 µg/kg Pb (the FDA candy/dried-fruit limit).
Methods (brief)
Analytical: ICP-MS (NexION 350X, PerkinElmer) in kinetic energy distribution (KED) mode using helium collision gas to reduce polyatomic interferences. Modified EPA method 6020A with EPA 3015A-based acid microwave digestion (HNO3 + HCl in water matrix; 10-min ramp to 170 °C, 10-min hold). LOQ: 2 µg/kg Cd, 4 µg/kg Pb. Method uncertainty: ±2.36% Cd, ±2.39% Pb. Calibration R² ≥ 0.998; RSD for matrix spike duplicates ≤ 20%. Lab: Ellipse Analytics, Denver, CO. Non-detects coded as 0 µg/kg for distributional analysis.
Sample selection: top national sellers by Nielsen 52-week US retail sales data, supplemented by natural/organic channel and direct-to-consumer brands. Purchased from Denver, CO area conventional and organic supermarkets, online retailers, and direct-to-consumer sources. Minimum 300 g per sample. Single sample per lot (not composite). Purchasing period not stated explicitly; paper received April 2018, accepted September 2018.
Statistical analysis: Wilcoxon rank-sum tests for ingredient associations (non-parametric, due to non-normal distribution); Kruskal-Wallis for product sub-category comparisons. SAS version 9.4.
Key limitation for HMT&C: Only Pb and Cd were measured; As, Hg, Ni, Al, Cr, Sn were not included. The study does not speciate As. The convenience sample is not randomized, and product sub-category n values are small (n=22 to n=140). Single-lot purchase means within-product lot variance is not captured.
Funding and independence: This study was funded by the Clean Label Project, an NGO. The corresponding author (Bowen) is affiliated with the Clean Label Project. The study uses peer-reviewed methods and is published in a reputable journal, but the organizational origin is relevant to evidence-tier interpretation. For HMT&C purposes, this study provides B-tier evidence at the product-category and ingredient-association level, but the raw concentration data (Table 1, Table 4) derived from ISO-accredited laboratory ICP-MS analysis are treated as A-tier observations.
Implications
Certification: This study is the primary source for the claim that 22% of US infant formulas exceed Proposition 65 lead limits at a 4-month-old’s recommended consumption level, and 14% exceed WHO PTMI for cadmium. These are the numbers brand regulatory and legal teams point to when evaluating whether certification to tighter standards adds defensibility value. For HMT&C, this study supports the argument that the current regulatory floor (FDA’s 6 µg/day Pb provisional limit for infant formula) does not protect against exposures that exceed stricter state or international benchmarks. It also demonstrates that organic certification is not a defensible proxy for low heavy metal content. The rice-association findings support HMT&C’s ingredient-based risk stratification approach.
Courses: The organic certification result (no significant difference in Pb or Cd vs. non-organic) is a high-impact teaching point: brands cannot claim low metal burden based on organic status alone. The dataset size (564 products) and the regulatory comparison framework across FDA, WHO, and Prop 65 make this paper valuable for explaining why certification programs that simply track one regulatory jurisdiction are insufficient.
App: The ingredient-association findings (rice, quinoa, wheat, oats elevating Cd; rice, sweet potatoes elevating Pb) are directly usable for ingredient-level risk scoring logic. The full distribution statistics (Table 1) provide a US-market reference distribution for Pb and Cd across all baby food sub-categories.