Parker 2022 — Arsenic, Cadmium, Lead, And Mercury In Baby Foods

Summary

This peer-reviewed study measured total arsenic, cadmium, mercury, and lead in 36 US baby and toddler food samples across four ingredient categories: fruit, grain, leguminous vegetable, and root vegetable. It is valuable for Category 1 because it reports n, detection frequency, minimum, mean, median, maximum, and standard deviation by matrix.

Study Scope

FieldValue
Product scopeBaby and toddler foods targeted to children from four months to three years
Total sample size36 products
Category sample size9 fruit, 9 grain, 9 leguminous vegetable, 9 root vegetable
Purchase frameFive supermarket chain locations in and around Pittsburgh, Pennsylvania, December 2018 to March 2019
Product typesOrganic and non-organic products in jars and pouches; three distinct lots per product type
Analytical methodICP-MS after heat-block assisted acid digestion, FDA EAM Method 4.7
Quantification rangeAs, Cd, Pb: 0.010 to 25.0 mg/kg; Hg: 0.010 to 2.50 mg/kg
LOD0.003 mg/kg, equivalent to 3 ug/kg, for As, Cd, Pb, and Hg
Non-detect handlingNon-detects assigned one-half LOD, 1.5 ug/kg, for risk assessment
Non-quantifiable handlingBetween LOD and LLOQ assigned one-half LLOQ, 5 ug/kg

Concentration Distribution By Ingredient Category

All values are in ug/kg, equivalent to ppb, from Table 4. Arsenic and mercury were measured as total elemental concentrations; speciation was not performed.

AnalyteCategoryNDetection frequencyMinimumMeanMedianMaximumSDTable
Total arsenicFruit96/9, 67%1.53.85.05.01.8Table 4
Total arsenicGrain99/9, 100%10.090.4126.0132.054.4Table 4
Total arsenicLeguminous vegetable97/9, 78%1.54.25.05.01.5Table 4
Total arsenicRoot vegetable99/9, 100%5.010.812.022.05.4Table 4
CadmiumFruit93/9, 33%1.54.41.516.05.2Table 4
CadmiumGrain99/9, 100%12.025.820.061.016.9Table 4
CadmiumLeguminous vegetable90/9, 0%1.51.51.51.50.0Table 4
CadmiumRoot vegetable96/9, 67%1.53.85.05.01.8Table 4
Total mercuryFruit90/9, 0%1.51.51.51.50.0Table 4
Total mercuryGrain90/9, 0%1.51.51.51.50.0Table 4
Total mercuryLeguminous vegetable90/9, 0%1.51.51.51.50.0Table 4
Total mercuryRoot vegetable90/9, 0%1.51.51.51.50.0Table 4
LeadFruit93/9, 33%1.52.71.55.01.8Table 4
LeadGrain99/9, 100%5.09.75.020.07.0Table 4
LeadLeguminous vegetable92/9, 22%1.52.31.55.01.5Table 4
LeadRoot vegetable98/9, 88%1.515.85.048.015.6Table 4

Distribution Limits

The authors state that each ingredient category had small sample size, N = 9, and that most samples across elements and categories were below the LOD. They therefore did not fit distributions, did not calculate a valid 95% upper confidence limit, and used maximum concentrations as upper-bound exposure concentrations. The public wiki should not infer p90 or p95 values from this study.

Risk Findings

FindingSource detail
Arsenic non-cancer HQs exceeded 1.0 for grain products across age groups under mean, median, and maximum concentration assumptions.Section 3.2
Arsenic cancer risk was driven by grain products, and the authors note that two of the three grain products were rice-based.Discussion
Cadmium and mercury did not produce expected non-cancer risks for any age group even at maximum concentrations.Discussion
Lead non-cancer risk signals were observed for fruit, grain, and root vegetable categories under selected exposure assumptions, with root vegetable and grain products being the more relevant categories.Sections 3.2 and Discussion
The highest predicted daily lead dose from baby food was 1.77 ug/day, below FDA’s interim daily intake level discussed by the authors.Discussion

Limitations

This study measured total arsenic and total mercury, not inorganic arsenic or methylmercury. It had only nine samples per ingredient category, and many values were censored by LOD or LLOQ substitution. The authors explicitly state that the data were insufficient to fit a distribution or calculate a valid 95% upper confidence limit.

Implications

  • Certification: Useful A-tier matrix-specific concentration evidence for grain, root vegetable, fruit, and leguminous vegetable baby foods, but not a source for p90 or p95 values.
  • Courses: Strong example of transparent LOD/LLOQ handling and why maximum values are not the same thing as percentiles.
  • App: Supports category risk flags for rice/grain and root vegetable rows, with total-As/speciation caveats.
  • Microbiome: No direct microbiome endpoint.

Wiki pages updated on ingest