Discussion
In our study, infants who were fed exclusively with breast milk had lower exposures to arsenic than those fed exclusively with formula or a mix of formula and breast milk, as determined by both urinary biomarkers and exposure modeling. Moreover, urinary arsenic increased with formula consumption and decreased with minutes of breastfeeding among infants who were not exclusively breastfed. These findings are consistent with the lower median concentration of arsenic measured in breast milk (0.31 μg/L) in the NHBCS compared with the combined concentrations of arsenic in formula powder (median 1.1 μg/L) and tap water (median 0.44 μg/L).
Our finding of low arsenic in breast milk was consistent with a previous study in Uppsala, Sweden, where drinking-water arsenic is low (Björklund et al. 2012). Although our estimate should be interpreted with caution given the small sample size (n = 9), studies consistently document much lower arsenic in breast milk than in drinking water (Table 3). Together, these studies suggest that breastfeeding is likely to result in lower infant exposures to arsenic regardless of arsenic concentrations in drinking water. This finding is consistent with studies of other metals such as lead (Ettinger et al. 2004; Gulson et al. 2001) and demonstrates an important public health benefit of breastfeeding.
Our exposure models further suggest that formula powder can make a large contribution to arsenic exposure for formula-fed infants when the arsenic concentration in drinking water is low (< 1 μg/L). Specifically, formula powder accounted for 71% of median estimated exposure in the NHBCS, suggesting that the powdered component of formula, rather than the mixing water, may be the primary source of exposure for many of the formula-fed infants in this population. This finding may be of particular concern given that the predominant form of arsenic in formula powder appears to be the more toxic inorganic species (Jackson et al. 2012). Identifying the sources of arsenic in formula powder could help reduce exposure for formula-fed infants if alternatives are available in the production process, consistent with earlier calls for greater attention to contaminants in infant formula (e.g., Ljung et al. 2011).
Median body weight–adjusted estimated arsenic exposures for NHBCS infants, assuming exclusive breast- and formula-feeding (0.04 and 0.22 μg/kg/day, respectively), were slightly lower than recent central-tendency estimates for European infants (mean, 0.24–0.43 μg/kg/day; EFSA 2014) and did not exceed the provisional tolerable weekly intake (PTWI) for arsenic previously used by the World Health Organization (15 μg/kg/week, or 2.1 μg/kg/day). Importantly, however, the WHO PTWI was withdrawn in 2010 as being insufficient to protect health, and recommendations have been made to reduce arsenic ingestion especially in young children (EFSA 2009, 2014). Therefore, our results reinforce recommendations for families with private wells to test for arsenic in their tap water and seek remediation if necessary.
Urinary arsenic concentrations in our U.S. infant population were lower than what has been observed in a highly exposed population in Bangladesh, even among breastfed children (Fängström et al. 2008). Mean urinary specific gravity in our infant subsample (1.003) was identical to that of 3-month-old Bangladeshi infants (Fängström et al. 2008), but lower than that of 18-month-old toddlers (1.009) (Hamadani et al. 2010). After adjustment to the average specific gravity, the median concentration of urinary arsenic in infant urine from the NHBCS (0.18 μg/L) was 6.5 times lower than 3-month-old Bangladeshi infants (1.2 μg/L) (Fängström et al. 2008). To our knowledge, no other study has investigated urinary arsenic in infants at 6 weeks of age from any region of the world, or breast milk arsenic in a U.S. population.
Limitations of our study include the lack of individual-level data for making individualized exposure estimates, the relatively small number of formula-fed infants, and the procedures for collecting breast milk. We were unable to calculate individualized exposure estimates for four reasons. First, we lacked body weight data. Second, we lacked data on ounces of milk ingested during breastfeeding events. Mothers reported the number of minutes on the breast, but we determined that extrapolation of minutes into ounces would be problematic due to the large variation in milk output both between and within individuals. Third, estimates of breast milk arsenic were available for only nine infants, not all of whom participated in our feeding substudy. Finally, although we have individual-level data on home tap water, our estimates rely on previous studies for data on arsenic in formula powder (Jackson et al. 2012) and bottled water (Sullivan and Leavey 2011). Although we had a fairly small sample of formula-fed babies, the mean formula ingestion rate from our population (0.81 L/day) is consistent with feeding recommendations from the American Academy of Pediatrics, Committee on Practice and Ambulatory Medicine, Bright Futures Periodicity Schedule Workgroup (2014) and the mean body weight–adjusted ingestion rate from our population is identical to the value recommended by the U.S. EPA (2008) for breast milk. Finally, breast milk was collected into the containers typically used by each parent, including plastic bags and bottles. Low levels of arsenic in these containers may have leached into breast milk and thus would overestimate exposure for infants fed milk directly from the breast rather than pumped. However, because 39% of parents reported feeding pumped breast milk, this was considered a conservative assumption, especially because arsenic contamination of plastic has not been reported in the literature.
We expect that population-wide arsenic exposure will increase during the second part of the first year of life, as the prevalence of formula-feeding increases and as solid foods are introduced. For example, rice, rice cereal, and common infant foods containing rice as a thickening agent can contain elevated concentrations of arsenic (Abedin et al. 2002; Hernández-Martínez and Navarro-Blasco 2013; Jackson et al. 2012; Meharg 2004). Also, rice has been shown to contribute to arsenic exposure in older children (Davis et al. 2012) and pregnant women (Gilbert-Diamond et al. 2011).
In conclusion, our findings suggest that breastfed infants have lower exposure to arsenic than formula-fed infants, even when drinking-water arsenic concentrations are low (< 1 μg/L). Moreover, our estimates suggest that both formula powder and drinking water can be sources of arsenic exposure for U.S. infants.