Overview
This Environmental Health Perspectives paper from the Children’s Environmental Health and Disease Prevention Research Center at Dartmouth (Carignan, Cottingham, Jackson, Punshon, Folt, Karagas) plus the University of Arizona (Gandolfi) reports estimated arsenic exposure in a U.S. infant cohort comparing breastfed and formula-fed infants at approximately 6 weeks of age. The paper uses urinary arsenic biomarkers paired with deterministic dietary intake modeling to characterize per-feeding-mode infant arsenic exposure within the New Hampshire Birth Cohort Study (NHBCS), a prospective cohort enrolling mother-infant pairs from private well-water households in New Hampshire. The study population consists of 72 infants from the NHBCS infant feeding substudy, drawn from a larger cohort of 937 participants with demographic data. Arsenic was measured in home tap water (n = 874 for the full cohort; n = 70 for the substudy), infant urine at 6 weeks (n = 72), and maternal breast milk (n = 9). The Dartmouth group’s NHBCS is the canonical U.S. prospective infant-arsenic-exposure dataset; this 2015 paper, the related Carignan 2016 commentary, and the Carignan 2016 first-year cohort paper form the U.S. infant arsenic prospective evidence cluster.
Study design
Prospective cohort substudy nested within the NHBCS. Enrollment began January 2009; recruitment for the infant feeding substudy ran July 2012 to April 2013. Eligibility required use of a private, unregulated well at the home occupied since the last menstrual period and plans to remain in that residence through delivery. Of 136 collection packets mailed, 97 (71%) were returned; 84 had a completed 3-day food diary and 82 had sufficient urine volume, yielding a final substudy n of 72. Three-day food diaries recorded feeding type, volume, breastfeeding duration, water source for formula reconstitution, and water source type for each feeding. Infants were assigned to one of three feeding categories based on two full diary days: exclusively breastfed (n = 48), exclusively formula-fed (n = 11), or mixed (n = 13). Among the feeding substudy households, 70% used home tap water to prepare formula more than half the time; 30% used bottled water more than half the time. The study was approved by the Committee for the Protection of Human Subjects at Dartmouth College; all participants provided written informed consent.
Analytes
Total arsenic and arsenic speciation were measured in urine. Species quantified: inorganic arsenic (arsenite, AsIII, and arsenate, AsV), monomethylarsonic acid (MMA), dimethylarsinic acid (DMA), arsenobetaine (AsB), and arsenocholine. Urinary arsenic used in analyses was defined as total arsenic minus arsenobetaine (AsB is considered nontoxic; arsenocholine was not detected). Total arsenic was measured in home tap water and breast milk. All arsenic in home tap water was assumed to be inorganic based on prior literature.
Instrument and analytical methods
Tap water was analyzed at the Dartmouth Trace Element Analysis (TEA) Core by ICP-MS using a quadrupole collision cell instrument (Agilent 7500c Octopole Reaction System) with helium as a collision gas to remove polyatomic interferences. Samples were acidified with Optima-grade HNO3 to 0.5% vol/vol. Calibration used NIST-traceable standards (NIST SRM 1640a for water). Analytical uncertainty was plus or minus 3 to 5% using weighted linear regression.
Infant urine was analyzed at the Arizona Laboratory for Emerging Contaminants (University of Arizona). Urine was thawed, vortexed, filtered at 0.45 micrometers, and diluted 10-fold in 10 mM ammonium carbonate with 5 mM EDTA. Total urinary arsenic was measured by ICP-MS with an external calibration in synthetic urine matrix using germanium as an internal standard. Arsenic speciation used HPLC-ICP-MS (Dionex GP50 pump, Hamilton PRP-100X column, collision cell ICP-MS) with a gradient mobile phase of 10 to 50 mM ammonium carbonate at pH 8.5 over 13 minutes; detection at m/z 75. Sample volume was 100 microliters at 1 mL/min column flow rate. Quality control included eight duplicate and three composite control samples.
Breast milk was analyzed at the Dartmouth TEA Core by ICP-MS (Agilent 7700x). One milliliter of breast milk was weighed into a Teflon vessel, 1 mL of 9:1 HNO3:HCl (Optima grade) was added, and the vessel was microwave-digested at 200 degrees Celsius for 20 minutes (MARS6, CEM Corporation), then diluted to approximately 10 mL. Calibration followed U.S. EPA (2007) Method 6020A initial and continuing calibration check procedures.
LOD/LOQ
Limit of detection for total urinary arsenic: 0.05 micrograms per liter. Tap water detection limit ranged from 0.009 to 0.074 micrograms per liter (mean, 0.014 micrograms per liter). The breast milk LOD is not reported as a single value; the paper notes arsenic in NIST SRM 1849a was measured at 4 ng/g, described as near the method detection limit. Detection limits for individual urinary arsenic species are not reported as single values; the paper notes that relatively high detection limits for individual species, combined with low total arsenic concentrations, explain the low detection frequencies for individual species. Urine values below the LOD for total arsenic were assigned a uniform random variate between zero and the LOD (Helsel 1990 method).
CRM and recovery
Tap water: NIST SRM 1640a used for standardization and reproducibility. Breast milk: NIST SRM 1849a included for quality control; selenium recovery in that SRM was 105% (selenium used as matrix-effect surrogate because arsenic is not certified in that SRM); arsenic spike recovery (n = 2) was 112%; fortified blank recovery 98%; duplicate relative percent difference 13%. No CRM recovery is reported for the urine matrix; quality control used eight duplicate samples and three composite controls.
Results
Measured arsenic concentrations
The following values are from Table 2 of the paper. Urinary arsenic is defined as total arsenic minus arsenobetaine; tap water data are log-normally distributed. Breast milk concentrations assume 1:1 conversion from ng/g to micrograms per liter following Bjorklund et al. (2012).
| Matrix | n | LOD (micrograms per liter) | Percent detected | Minimum | 25th percentile | Median | 75th percentile | Maximum |
|---|---|---|---|---|---|---|---|---|
| Home tap water, full cohort | 874 | 0.01 | 84% (736) | below LOD | 0.07 | 0.44 | 2.72 | 189 |
| Home tap water, substudy | 70 | 0.01 | 71% (50) | below LOD | 0.03 | 0.15 | 0.57 | 29 |
| Infant urine | 72 | 0.05 | 97% (70) | below LOD | 0.07 | 0.17 | 0.37 | 2.9 |
| Breast milk | 9 | 0.22 | 67% (6) | below LOD | 0.25 | 0.31 | 0.44 | 0.62 |
Urinary arsenic by feeding mode
Geometric mean urinary arsenic (adjusted for specific gravity) was 7.5 times higher in exclusively formula-fed infants than in exclusively breastfed infants (beta = 2.02; 95% CI: 1.21 to 2.83; p less than 0.0001). Mixed-fed infants had urinary arsenic approximately 3 times higher than exclusively breastfed infants (beta = 1.08; 95% CI: 0.34 to 1.83; p = 0.005). The difference between exclusively formula-fed and mixed-fed infants was marginally significant (beta = 0.94; 95% CI: minus 0.04 to 1.91; p = 0.06).
Among infants not exclusively breastfed, each additional ounce per day of formula was associated with a 2.6% increase in urinary arsenic (beta = 0.026; 95% CI: 0.010 to 0.042; p = 0.003). Minutes of breastfeeding per day were inversely associated with urinary arsenic in that group (beta = minus 0.004; 95% CI: minus 0.006 to minus 0.002; p = 0.001). Specific gravity adjustment was applied as a covariate. Mean specific gravity was 1.003 plus or minus 0.0014.
Arsenic speciation in urine
Detection frequencies overall: arsenobetaine (AsB) 19%, inorganic arsenic 24%, MMA 8%, DMA 38%. DMA was detected in urine from all 11 formula-fed infants versus 9 of 48 breastfed infants (Fisher’s exact test, p less than 0.0001). MMA was detected in 3 of 11 formula-fed infants and in none of 48 breastfed infants (Fisher’s exact test, p = 0.005). Detection frequencies for inorganic arsenic and AsB did not differ by feeding mode (both p greater than 0.4).
Exposure modeling results
The deterministic exposure model used a standardized ingestion rate from exclusively formula-fed infants in the diary substudy (mean 0.81 L/day; range 0.58 to 1.18 L/day) and body weight from U.S. EPA Child-Specific Exposure Factors Handbook data for 1- to 3-month-old infants (mean 5.6 kg; range 4.5 to 7.3 kg). Arsenic in formula powder was taken from the Jackson et al. (2012) NHBCS market basket study (15 formulas from 5 brands; median 1.1 micrograms per liter; range 0.3 to 1.8 micrograms per liter). Minimum and maximum exposure estimates were paired by ingestion rate and body weight because of their underlying correlation.
Modeled median daily arsenic exposures:
| Feeding mode / water source | Median estimated exposure (micrograms per kilogram per day) | Maximum estimated exposure (micrograms per kilogram per day) |
|---|---|---|
| Exclusively breastfed (NHBCS breast milk, median 0.31 micrograms per liter) | 0.04 | n/a |
| Exclusively formula-fed, NHBCS tap water (median 0.44 micrograms per liter) | 0.22 | 31 |
| Exclusively formula-fed, bottled water (median 0.62 micrograms per liter) | 0.25 | 0.60 |
Formula powder accounted for approximately 71% of median estimated exposure for formula-fed NHBCS infants (at median tap water arsenic of 0.44 micrograms per liter). The proportion attributable to formula powder decreased with increasing tap water arsenic: approximately 52% at 1 microgram per liter, declining to approximately 10% at 10 micrograms per liter. The maximum estimated exposure of 31 micrograms per kilogram per day was driven almost entirely by a single household with tap water arsenic at 189 micrograms per liter.
The 5.5-fold median exposure difference between formula-fed (0.22 micrograms per kilogram per day) and breastfed infants (0.04 micrograms per kilogram per day) is consistent with breast milk median arsenic (0.31 micrograms per liter) being lower than the combined formula powder plus tap water arsenic. These median estimated exposures are slightly lower than European central-tendency estimates for infants (mean, 0.24 to 0.43 micrograms per kilogram per day; EFSA 2014). The WHO provisional tolerable weekly intake (PTWI) for arsenic (15 micrograms per kilogram per week, equivalent to 2.1 micrograms per kilogram per day) was not exceeded at median estimates; the WHO withdrew the PTWI in 2010 as insufficient to protect health.
Evidence fitness
This paper provides strong urinary biomarker evidence for a feeding-mode effect on infant arsenic body burden in a U.S. prospective cohort, and a deterministic exposure model with well-characterized input distributions. The urinary biomarker findings are based on 72 infants (48 breastfed, 11 exclusively formula-fed, 13 mixed), with the formula-fed subgroup relatively small. The breast milk subsample is nine mothers, limiting precision of the breast milk arsenic estimate. The arsenic concentration in formula powder is not directly measured in this study; the exposure model relies on published data from the Jackson et al. (2012) NHBCS market basket study for formula powder and from Sullivan and Leavey (2011) for bottled water. The study population is drawn from households with private unregulated wells in New Hampshire, which is not nationally representative; tap water arsenic in the substudy was substantially lower than in the full NHBCS cohort. These limitations are explicitly acknowledged by the authors. The key quantitative findings (the 7.5-fold urinary arsenic difference by feeding mode and the 5.5-fold modeled exposure difference) are internally consistent and consistent with the broader literature on breast milk arsenic being lower than formula-plus-water arsenic.
Limitations
Explicitly noted by the authors: (1) individualized exposure estimates were not possible because individual body weight data were not available, ounces ingested during breastfeeding could not be reliably extrapolated from minutes, and breast milk arsenic was available for only 9 of the 72 substudy infants; (2) the formula powder arsenic estimate relies on Jackson et al. (2012), not on direct measurement within the feeding substudy; (3) the formula-fed subgroup is small (n = 11); (4) breast milk was collected into the containers typically used by each parent, including plastic bags and bottles, which may introduce low-level arsenic contamination — the authors note this is a conservative assumption because 39% of parents reported feeding pumped breast milk and arsenic contamination of plastic had not been reported in the literature at the time of publication; (5) the substudy population differs from the full NHBCS in tap water arsenic distribution (chi-squared = 9.53, p = 0.01), with significantly fewer high-arsenic households in the substudy.
Regulatory context
The U.S. EPA maximum contaminant level (MCL) for arsenic in public drinking water is 10 micrograms per liter (U.S. EPA 2001). Private well water is not regulated under the MCL. In the full NHBCS, approximately 10% of wells exceeded the MCL; in the feeding substudy subset, only 1% exceeded the MCL. The New Jersey MCL for arsenic in drinking water is 5 micrograms per liter (New Jersey Administrative Code 7:10, 2011). The WHO provisional tolerable weekly intake for arsenic (15 micrograms per kilogram per week, 2.1 micrograms per kilogram per day) was withdrawn in 2010. EFSA (2014) reported central-tendency dietary arsenic exposure estimates for European infants in the range of 0.24 to 0.43 micrograms per kilogram per day, slightly above the NHBCS median formula-fed estimate of 0.22 micrograms per kilogram per day reported here.
Key conclusions
Formula-fed infants had urinary arsenic 7.5 times higher than exclusively breastfed infants at 6 weeks of age in this U.S. private-well household cohort. Modeled median daily arsenic exposure was 5.5 times higher for formula-fed infants (0.22 micrograms per kilogram per day) than for breastfed infants (0.04 micrograms per kilogram per day). At the median NHBCS tap water arsenic concentration, formula powder accounts for approximately 71% of median modeled arsenic exposure for formula-fed infants. Both formula powder and drinking water are sources of arsenic exposure; the relative contribution of formula powder is highest when tap water arsenic is low and decreases as tap water arsenic rises. Breastfeeding appears protective against arsenic exposure even in this predominantly low-arsenic-water population.
Implications
- Certification: Direct primary-evidence input for the per-feeding-mode arsenic exposure profile relevant to HMTc infant formula row scoping. The breastfed-vs-formula-fed comparison provides quantitative support for the infant formula heavy metal load concern that the certification-floor case rests on. The finding that formula powder accounts for approximately 71% of median exposure when tap water arsenic is low (0.44 micrograms per liter) reinforces that formula powder arsenic content, not reconstitution water alone, is the primary exposure driver for most U.S. households.
- Courses: Standard reference for U.S. prospective infant arsenic exposure data. The feeding-mode comparison and exposure model are teachable because the inputs are publicly reported and the arithmetic is deterministic.
- App: Supports the formula-vs-breastmilk per-serving arsenic exposure estimate. The 0.22 vs 0.04 microgram per kilogram per day median figures are suitable as reference estimates for a formula versus breastmilk exposure comparison tool.
Related evidence
- Arsenic, Inorganic
- Arsenic, Total
- Infant Formula, Powder (Non-Soy)
- Infant Formula, Powder (Soy-Based)
- Arsenic concentration and speciation in infant formulas and first foods
- Contribution of breast milk and formula to arsenic exposure during the first year of life in a U.S. prospective cohort
- Potential Exposure to Arsenic from Infant Rice Cereal
Update history
No substantive edit history is available in this build. The full commit record is available in git.