Dietary Exposure to Toxic Elements and the Health of Young Children: Methodological Considerations and Data Needs

Abstract

Concerns have been raised regarding toxic-element (arsenic, cadmium, lead, and mercury) contamination of commercially available infant foods around the world. Young children are vulnerable to the effects of toxic elements, based on higher absorption levels and potentially poorer detoxification capacities. Toxic-element exposures in early life exact high societal costs, but it is unclear how much dietary exposure to these elements contributes to adverse health outcomes. Well-designed epidemiological studies conducted in different geographical and socioeconomic contexts need to estimate dietary toxicant exposure in young children and to determine whether causal links exist between toxicants in children’s diets and health outcomes. This commentary outlines the methodological considerations and data needs to advance such research.

Many infants are introduced to complementary foods by 4–6 months of age (1, 2, 3), even if human milk or formula feeding continues longer. Concerns have been raised regarding the presence of nonessential, toxic elements (arsenic, cadmium, lead, and mercury) in infant foods sold in the United States (4, 5, 6) and countries around the world (7, 8, 9). Infant formula is another source of exposure, with a range of toxic-element levels reported across countries (10, 11, 12, 13). Metal content in foods varies by factors related to food production (14) including cultivars; seasons; soil and water usage and contamination; air pollution and dust; animal feed type; and food transport, processing, and preparation (15, 16, 17, 18), including packaging, technological processes, and cookware. Consequently, commercial and homemade foods could contribute to dietary exposure to toxic elements in young children. We specifically focus on the nonessential toxic elements of arsenic, lead, mercury, and cadmium because they rank first, second, third, and seventh, respectively, on the National Priorities List of the US Agency for Toxic Substances and Disease Registry (19).

We define dietary exposure to these toxic elements (Figure 1) as exposure that occurs through the ingestion of foods, beverages, and supplements that have been contaminated during production, transport, storage, or preparation and cooking. We include water used for cooking and drinking, as well as infant formula. It is also important to acknowledge that human milk may be a source of toxic elements (20) for young infants and is, therefore, part of the dietary exposure definition.

FIGURE 1.

Definition of dietary exposures to toxic elements. *Water used for cooking and drinking is included.

In the United States, the FDA released guidance on inorganic arsenic in rice cereals (21), Consumer Reports cautioned parents about lead and arsenic in fruit juice (22), and a study of 168 foods conducted by Healthy Babies Bright Futures detected arsenic in 73% of baby foods, cadmium in 75%, lead in 94%, and mercury in 32%, with all 4 metals found in 26% of foods tested (4). International agencies, like the FAO, WHO, and the European Food Safety Authority (EFSA), have established limits for toxic elements in foods (23, 24, 25). Specific countries like China and Brazil also have comprehensive guidelines (26, 27). The true extent of contamination of foods consumed by young children globally is unknown and may vary regionally due to differences in crop cultivation and livestock feeding conditions. Recently, the US FDA launched the Closer to Zero (C2Z) Action Plan to reduce contamination in infant foods (28); in support of this plan, the interim reference level for dietary lead in the United States was updated to 2.2 μg/d (29). Nevertheless, the multiphase plan proposed by the FDA has protracted timelines (e.g., beyond 2024 for arsenic in “foods consumed by babies and young children”) and may not lead to timely regulatory action. Calls for closer testing at various stages of food production and for the use of warning labels have been issued (30). Commercially produced foods (finished products) in the United States are not systematically monitored by manufacturers for toxic-element content. Monitoring of homemade baby foods that use ingredients available to the general population and not specifically labeled for infants and young children would be very difficult.

Children consume higher amounts of contaminants per body weight (6) and absorb higher proportions of elements compared to adults. For example, proportional lead absorption in adults ranges from 8% to 35% depending on the fasting status (31), but average lead absorption in infants can reach 41% with dietary lead intake > 5 μg/kg/day (32) and may increase with iron deficiency (33, 34). Young children may also have underdeveloped mechanisms for detoxifying toxic elements (35). Considering developmental trajectories, even low- and moderate-level exposures to toxicants during early childhood may impact early life and later health and development (36).

Societal costs of toxic-element exposures are substantial (37, 38, 39). Table 1 briefly describes 2 types of studies: those that link diets or foods with biomarkers of toxic-element exposure in young children and those that investigate the associations between toxic elements in the postnatal period and their relations to children’s health. It is important to clarify that studies on health effects are mostly based on biomarkers of exposure, which integrate across all sources, not only diet. Few studies have systematically investigated the extent to which exposures through typical diets contribute to internal markers of toxic-element exposure or to adverse health outcomes in young children. One study reported that the global impact of dietary lead on the severity of intellectual disability was low but not negligible (40), whereas another estimated that 36.6 per 1000 healthy life-years are lost among 0- to 6-year-old Chinese children (41). Dietary exposure may have important public health implications, as shown for rice, a common source of toxic forms of arsenic (42). Both prenatal arsenic exposure and early introduction of rice cereal were recently linked with higher risks of infections and respiratory symptoms among US infants (43, 44). Another study estimated the noncancer hazard index scores from arsenic and lead ingestion from baby foods to be >1 (indicating possible health effects) among US children aged 0–3 years and 1–3 years, respectively. This study also identified potential cancer risks related to arsenic consumption (45).

TABLE 1.

Select studies demonstrating the relationships between children’s diets and food consumption and the biomarkers of arsenic, cadmium, lead, and mercury, as well as observed health outcomes in association with toxic-element exposure in early childhood¹

Toxic element	Biomarkers	Neurobehavioral development	Other outcomes
Arsenic	Infants from New Hampshire had higher inorganic arsenic and higher concentrations of methylated arsenic species in urine during weaning than when exclusively fed breast milk, formula, or both. Changes in urinary arsenic concentrations were linked to consumption of rice cereal, fruit, and vegetables (80)	In Spanish children aged 4–5 years, the concentration of the sum of arsenic species in urine (median 4.85 μg·L−1) was negatively associated with fine and gross motor functions in all children and with quantitative index and working memory test performance in boys (81)	In 6-week-old infants from New Hampshire, urinary arsenic concentrations were associated with fecal enrichment of 8 microbial genera, 6 in the phylum Firmicutes. Conversely, 15 genera were negatively associated urinary arsenic, including Bacteroides and Bifidobacterium. These findings were more pronounced in formula-fed, male infants (83)
		Water arsenic levels in households of children aged 20–40 months from Bangladesh were associated with lower cognitive scores measured at the same ages (82)	4- to 6-year-old children from Bangladesh with arsenic in drinking water > 50 μg·L−1 had a higher relative abundance of phylum Proteobacteria fecal DNA than children with water arsenic < 10 μg·L−1. There was some evidence that higher arsenic exposure may be related to antibiotic resistance (84)
Cadmium	Using a dietary-wide association study of 49 foods, a study of children participating in the US NHANES estimated that diet explains 1.4% of variability in blood (biomarker of shorter-term exposure) and 1% of variability in urinary (long-term exposure biomarker) cadmium concentrations (85)	Urinary cadmium at 5 years (median, 0.22 μg·L−1) was negatively associated with the performance and full-scale IQ at the same age, independent of pregnancy exposure (86)	Dietary cadmium exposure between 1 and 9 years of age (mean intake, 4.43–8.09 μg·d−1) that exceeded tolerable weekly intake levels across childhood was associated with a lower blood urea nitrogen level at 9 years but not more specific markers of kidney function (87)
			Urinary cadmium at 4–5 years (median, 0.3 μg·g−1 creatinine) was negatively associated with standing height measured concurrently in Spanish children (88)
Lead	Higher BLLs were found among 12- to 36-month-old US children participating in the NHANES who consumed higher amounts of apple juice; lower concentrations were observed among children with higher intakes of breakfast cereals (89)	The BLL at 24 months of age (mean < 5 μg·dL−1) was associated with lower mental development scores in Chinese children at 24 and 36 months (91)	BLLs (median, 0.663 μg·dL−1) were measured in Canadian children aged 22–48+ months. The highest tertile of BLLs was associated with lower BMI-for-age z-scores among girls (93)
	Diet made the strongest estimated contribution to the BLL among 1- to 2-year-old US children with BLLs below 1.5 μg·dL−1 (90)	Hair lead concentrations (mean, 0.8 ± 5.1 μg·g−1) were negatively associated with expressive language scores in children < 3 years of age from Taiwan (92)	Among Chinese children (mean age, 4.85 years; median BLL, 4.8 μg·L−1), interquartile changes in BLL were associated with 8%–30% lower concentrations of serum neurotransmitters glutamic acid, gamma-aminobutyric acid, and glycine (94)
		BLLs (median, 7.6 μg·dL−1) at 20–40 months were associated with lower cognitive scores measured concurrently in children from Bangladesh (82)
Mercury	Fish intake occurring more than once per week over a typical month was associated with higher hair mercury content among Korean infants aged 6–24 months (95)	Hair (mean, 1.96 μg·g−1) fingernail (mean, 0.64 μg·g−1) and toenail (mean, 0.55 μg·g−1) mercury concentrations at 3 years were negatively associated with expressive language performance measured concurrently in Taiwanese children (97)	Hair mercury (geometric mean, 0.97 μg·g−1) in 4-year-old Spanish children was not associated with respiratory outcomes in cross-sectional analyses (100)
	Mother-child pairs from 17 European countries with high consumption of sea fish, shellfish, seafood products, and freshwater fish had higher hair mercury concentrations than pairs with lower levels of consumption (96)	Urinary mercury concentrations at 3–12 months were negatively associated with nonverbal IQ and visual-motor ability integration at 5–8 years (98)
		Methylmercury concentrations in hair (geometric mean, 0.98 μg·g−1) at 4 years were positively associated with Spanish children’s verbal test performance, but there was uncertainty regarding assessments of fish intake and mercury exposure (99)

¹BLL, blood lead level.

Epidemiological studies are needed to estimate the levels of toxicant exposure in young children that occur from the diet. They also need to determine whether causal links exist between toxicants in children’s diets and biomarkers of the internal dose of exposure and with health outcomes (Figure 2). The objective of this perspective is to outline the methodological considerations and data needs to advance such research. We focus on dietary exposures during childhood, but our framework applies to prenatal exposures, which are included within the first 1000 days of life. While we recognize the potential toxicity of some elements, like fluoride (46, 47), their role as essential nutrients means that they deserve a separate discussion in the literature.

FIGURE 2.

Conceptual framework for the relationship between dietary exposure to toxic elements, internal dose of exposure, and child health and development. ¹Dietary nutrients may serve as absorption inhibitors or as detoxifiers, thus contributing to effect modification. ²Dietary nutrients may independently promote child health or modify the association between exposure and health. ³Includes natural and processed foods and beverages, with contamination occurring during cultivation, transport, processing, storage, or preparation and cooking. Dashed lines represent relationships that have been hypothesized or remain to be elucidated. Solid lines represent established relationships. Lines with a stop symbol indicate effect modifications; arrows indicate main effects.

Dietary assessment

Accurate dietary assessment is a critical component of exposure and risk assessment; it helps determine the level of toxicant exposure via childhood diets and the proportion of the population for whom dietary exposure exceeds established reference values. Dietary assessments should be combined with testing of toxic elements in representative market baskets of foods, once prepared for consumption. Numerous surveys measure dietary intakes of US children, including the NHANES and the Feeding Infants and Toddlers Study (FITS) (48). Surveys such as NHANES and FITS rely on a 24-hour multipass method of recall to achieve greater accuracy. A single 24-hour recall is suitable to estimate population-level nutrient intakes (49); exceedance of toxicant reference values by individual children should not be determined based on a single recall. Other surveys, such as Mexico’s National Survey on Health and Nutrition (ENSANUT), use both a 24-hour recall and a semiquantitative FFQ for children aged 1–4 years.

The assessment of infant diets poses special challenges for accurate exposure assessment of toxic elements. Diets change frequently, both with seasons and as infants age, and diets vary with geographical area, culture, and ethnicity. A single assessment (e.g., at 12 months) is unlikely to represent the diet (or dietary toxicant exposure) throughout infancy. Methodologically, 24-hour recalls collect the amounts of food intake, but multiple recalls are needed to represent typical intakes, especially for seasonal foods. FFQs are excellent alternatives if they incorporate information on serving sizes and are validated against adequate nutritional biomarkers (50, 51), as well as toxicant biomarkers.

Toxic elements have been detected in human breastmilk but there is large variability in observed levels (20, 52, 53). Formula may make a greater contribution to exposure (54). It is difficult to quantify the amount of breastmilk intake and, by extension, the level of dietary exposure to toxicants in breastfed infants. The quantification of expressed or donated milk intake is easier by comparison. With respect to breastfeeding, querying mothers on the frequency and duration of feeding episodes does not fully reflect the amount consumed, given the variations in milk production and in infant activity. Methods comparing pre- and postfeed weights are considered imprecise (55). A more robust technique uses stable isotopes, such as deuterium oxide, which are repeatedly sampled in maternal and infant saliva (56). Human milk intake can then be estimated using standardized calculations (57, 58). This method has been used to assess the transfer of pesticides from breastmilk to infants (59). While this method is accurate and noninvasive, it is impractical for large population studies, as it requires frequent sample collection and is expensive due to the cost of deuterium oxide.

Addressing data and methodological needs (Table 2) in several areas of dietary assessment would improve our ability to investigate the extent of dietary exposure to toxic elements and any associated health risks.

TABLE 2.

Data and methodological needs to advance research on the links between dietary exposure to toxic elements and child health¹

Dietary assessment

•Expand dietary assessment to younger ages. In the US NHANES, for example, there are small samples of young children for whom diet was assessed, particularly in those below 12 months of age (101). Collection of data at younger ages is relevant for capturing the transition to solid foods.

•Regularly update dietary databases to capture changes in the foods, beverages, and supplements available on the market and consumed by young children (101).

•Introduce or extend biomarker collection in national surveys and scientific studies. For example, no blood is collected from children younger than 12 months in the US NHANES, thus limiting the study of the relationship between diet and foods and biomarker levels. Furthermore, when biomarkers are only collected on a subsample of children, low sample sizes may limit the statistical power to detect associations.

•Consider the period of recall. To reflect dietary toxicant exposures, the reference period for recall should be short enough (2–4 weeks) to account for the changing nature of young-child diets, yet timed appropriately to the exposure period represented by any corresponding biomarkers (e.g., 1–2 days for urinary arsenic and 4–6 weeks for blood lead).

•Use validated instruments for the given geographical region and population group and, if possible, estimate the levels of under- and overreporting (9). The latter may be difficult for young children, who often consume snacks rather than meals, may only partially eat the foods served, or may have multiple caregivers feeding them.

•Include an assessment of water consumption from different sources (household tap, bottled, storage tanks, etc.) and, if possible, link information on the amount of intake with toxicant levels measured in water (71).

•Assess consumption of infant formula, including type and amount.

•Assess the intakes of dietary supplements, as they may contain toxic elements (102, 103, 104).

•Incorporate validated methods to measure human milk intake into the estimation of toxic element exposure and risk assessment.

Measuring toxic elements in foods

•Frequently and comprehensively assess toxicant concentrations in foods.

○ Recent data are critical, as contaminant levels in foods may change with seasons and vary over time and across regions, even for the same brands and products. Foods are periodically reformulated and discontinued, and new foods are introduced

○ For exposure and risk characterization among infants, more infant foods should be included. For example, while the US TDS includes infant foods, the number of infant foods is limited and other TDSs (e.g., sub-Saharan Africa) did not test any infant foods (60)

○ Foods from culturally and ethnically diverse diets need to be represented

○ The inclusion of imported foods should proportionally mirror their availability on the given market

○ Data from multiple studies could be consolidated but would require documentation and release of information on product names and brands, the date and location of sample collection, the laboratory methods, and LODs

•Expand assessment of contaminant types. Toxic elements and minerals, such as aluminum, chromium, fluoride, and nickel, occur in the diets of young children; dietary exposure to these elements early in the lifecycle may affect child growth and development (46, 105, 106).

○ Speciation analyses of elements with multiple stable chemical forms that differ on toxicity (e.g., arsenic and mercury) should be performed to accurately estimate exposures (24)

•Implement continuous human milk biomonitoring. Studies of toxic elements in human milk are needed, perhaps in parallel with monitoring of the persistent organic pollutants outlined in the Stockholm convention (107).

•Conduct research in special population groups. Information on dietary toxic-element exposure is lacking for vulnerable groups, including pregnant women; children with food allergies, metabolic diseases, or developmental disorders; or children on special diets. They may restrict or overconsume some types of foods that could put them at risk of nutritional deficiencies or high exposure to toxicants.

Laboratory analytical methods

•Improve comparability and data integration. To do this, laboratories need to move to ICP-MS for contaminant detection, develop and maintain validated methodology, work to lower their detection limits, and participate in quality control programs.

•Collaboration, knowledge, and technology transfers will facilitate the adoption of ICP-MS in lower-income countries.

Calculating metrics to reflect dietary exposure to toxic elements

•Given the large geographical variability in food contaminants, particularly for fresh foods, studies need to develop and use population-specific databases of contaminant values in foods.

•To the extent possible, studies need to employ probabilistic methods to estimate dietary exposure to toxic elements.

•Given the need to investigate mixture effects, and the fact that elements co-occur in foods, methods estimating joint dietary exposure to elements need to be developed and validated.

Causal inference models

•The association between dietary toxic-element exposure, biomarkers of exposure, and health effects in children should be investigated in diverse geographical settings and populations in large, longitudinal studies.

•When exposures occur as mixtures, their adverse effects on children’s health may be greater than the effects of the individual elements alone (108), even at low levels. Because diet is a source of multiple toxicants, epidemiological studies should, on the one hand, investigate the dietary predictors of elevated levels of toxic-element mixtures and, on the other hand, model mixtures as predictors of health outcomes.

•For models testing the associations between dietary toxicant exposure and biomarkers of exposure, adjusting exposure levels for proportional absorption or, alternatively, considering the level of food or nutrient intake from the diet may be important. The contribution of dietary exposure to biomarker levels could be estimated through accurate exposure assessment and calculation of the proportion of variance in biomarker explained by dietary exposure.

•Nutritional deficiencies may affect the association between dietary toxicant exposures and their biomarkers. For example, iron deficiency may increase the intestinal absorption of divalent cations like cadmium and lead (109).

•Causal models of the relationships between dietary toxicant exposure and health outcomes need to account for exposure from other sources (e.g., dust, soil, paint, air pollution, industrial activities, occupational exposures of household members).

○ Toxicant-specific sources can be determined from published literature and knowledge of local context

○ Studies need to consider that exposure from other sources could be time-varying

•Causal models of exposure-outcome relationships need to consider the effects of nutritional status and nutrient intakes on child health. The latter factor is critical, as diet is a mixture of toxicants and health-benefitting nutrients.

•Neurobehavioral development has long been the focal point of environmental epidemiology studies, given the importance of cognitive performance and behavior for scholastic achievement, but other health endpoints should also be considered, such as cardiometabolic outcomes (110) and neuroendocrine system function (111).

¹ICP-MS, inductively coupled plasma mass spectrometry; LOD, limit of detection; TDS, Total Diet Study.

Measuring toxic elements in foods

The contamination of the food supply in many countries around the world is monitored via location-specific Total Diet Studies (TDSs) (60, 61, 62, 63), with the foods selected for analysis typically representing 90%, by weight, of the average total diet (60). In the United States, select foods are purchased from grocery stores and fast-food restaurants (64). Foods are pooled, prepared (e.g., cooked), then analyzed. In the US TDS, currently available data combine the results over a nearly 15-year period (2003–2017). TDSs in other settings have followed similar methodologies adapted to local contexts (60). Additional data on specific foods are found in scientific publications, with methodologies for food collection, preparation, and laboratory analyses varying widely. A 2019 report by Healthy Babies Bright Futures (4) is currently among the largest compilations of data on toxic elements in baby foods sold in the US, including arsenic speciation. Generally, available data on food contaminants are incomplete and largely outdated.

As a potential source of toxic-element exposure, measurement of toxicants in milk may help estimate the total exposure doses for breastfed infants. However, it is important to recognize that the toxic-element content of human milk depends on the milk type (e.g., colostrum compared with mature milk), duration of lactation (20), and maternal (milk donor) factors, such as nutritional status, diet, and smoking (65), the woman’s exposure level, and bone turnover (66). There is some evidence that methylation or other mechanisms may lower the transfer of arsenic and mercury to the infant, while transfer of lead and cadmium is less impeded (20). The daily intake of these elements for many of the world’s infants is estimated to be low, but this is likely untrue in areas with higher environmental contamination (20) and, given the harms of toxic elements even at low levels of exposure, no assumptions of safety should be made.

Future studies should address several shortcomings with respect to contaminant assessment in foods (Table 2).

Laboratory analytical methods

To date, most studies have employed atomic absorption spectrometry (AAS) and inductively coupled plasma mass spectrometry (ICP-MS) to measure toxic-element contents of foods, formula, and breastmilk. In AAS, the element analyte is thermally decomposed to atoms that absorb light at a particular wavelength. The AAS has a relatively high limit of detection (LOD), which refers to the lowest analyte concentration at which detection is feasible. ICP-MS utilizes a high-energy argon plasma to convert the sample constituents to their elemental components, which are then ionized and transported to the mass spectrometer for selective detection and quantification. The ICP-MS can be coupled with chromatography to perform a speciation analysis, which is important for those elements which occur in foods in different forms, like arsenic and mercury (e.g., arsenobetaine and arsenocholine, the organic forms of arsenic, as well as dimethylarsinic acid (DMA), a methylated form of inorganic arsenic). These are considered to have different levels of toxicity in the human body (24), so understanding which species are present in foods is critical to understanding health risks. Currently, the ICP-MS is the gold-standard technique for trace element determination, due to many desirable features, such as low LODs, a multielement capability, and a wide linear dynamic range (24, 67). The difference in LODs between AAS and ICP-MS can be large and could result in exposure misclassifications, as well as could prevent the combining or comparing of data. Table 2 outlines research needs with respect to laboratory methods.

Calculating metrics to reflect dietary exposure to toxic elements

The most common method of combining dietary intake with contaminant levels is a deterministic calculation based on matching specific foods available in a dietary database with toxicant concentrations in corresponding foods from a food contaminant database, like those generated from TDSs. One challenge is that dietary databases tend to be more extensive than contaminant databases. For instance, 53 fruit purees or fruit-based baby foods appear in the US Food and Nutrient Database for Dietary Studies, which provides nutrient profiles for foods and beverages reported in the NHANES. In contrast, only 17 fruit-related foods are available in the US TDS, and only 4 are baby foods. Information on brand names may be absent or incomplete in 1 or both databases. This uneven matching means the following: 1) foods available in the contaminant database may be substituted for foods that are absent; 2) foods absent from 1 or both databases may be excluded from analysis altogether; and 3) matching of foods between databases may be driven by analyst knowledge or preferences. Decisions may be incorrect or arbitrary, leading to poorly reproducible findings, exposure misclassification, and, very likely, underestimation of exposure.

Probabilistic modeling has been developed to address issues in data variability and uncertainty, as well as the exposure underestimation that results from incomplete information (68, 69). While complex and time-consuming, probabilistic models consider the entire range of possible or available exposures and uncertainties inherent to the variables used in the analysis. Deterministic and probabilistic modeling approaches both have utility in exposure assessment and have been used both to estimate health risks of exposure (70) and in causal inference models to relate dietary exposures to biomarker levels (71) and child health outcomes (46). The accuracy of exposure metrics based on these models will improve with better data inputs, as discussed in prior sections. Table 2 outlines research needs for exposure estimation.

Causal inference modeling to understand the relationships between dietary exposures to toxicants and child health

Relationships between dietary exposures and biomarkers of exposure (Figure 2A) should be examined prior to studying the relationships between dietary toxicant exposures and child health (Figure 2B). However, there are at least 3 considerations regarding the exposure-biomarker relationship. First, only a portion of the dietary toxicant dose is absorbed by the intestinal epithelium. Proportional absorption depends on the amount of toxicant ingested, lengths of intervals between meals (72, 73), and bioavailability of the element to the intestinal epithelium, which is in turn influenced by the nutrients present in foods (74, 75, 76), intestinal microbiota (77), and the mode of consumption [e.g., types of foods served together, raw compared with cooked vegetables (78), etc.]. Table 2 outlines specific considerations for future causal inference studies in this arena.

Achieving critical data and evidence needs

There is a real need for larger, updated, more comprehensive data sets on dietary intakes and food contamination than are currently available. To achieve such advances, the cooperation of appropriate national-level agencies is required in each country. In the United States, this signifies working with the National Center for Health Statistics of the CDC and the FDA to expand data collection and coverage through the NHANES and the TDS, respectively. Furthermore, collaboration with the FAO, WHO, and EFSA to improve monitoring of the food supplies and dietary intakes would move the needle on data availability globally. There is guidance, for example, on harmonized methodology for the TDS (79). Academic centers and nonprofit organizations could also contribute by adopting harmonized survey and laboratory methods to complement or expand the data releases by national surveys, but some level of coordination may be required. Making the data publicly available for use in research is also critical to the advancement of knowledge. Without these investments in data, the science will remain piecemeal and there will be major delays in generating the necessary evidence.

In conclusion, epidemiological studies using age-appropriate dietary instruments, comprehensive and updated contaminant databases, and statistical (e.g., probabilistic exposure models, mixture modeling) approaches are necessary to improve our understanding of both dietary exposure to toxic elements and their health effects in young children, improve exposure monitoring, and inform regulatory decisions. Recognition of the importance of this type of translational research and dedicated funding streams would help in the timely generation of evidence. Creating partnerships between government and academia in regulatory programs like the C2Z in the United States can also aid in generating the evidence base (observational or interventional) for timely decision making. Finally, mechanisms for time-sensitive regulatory interventions will help to protect young children from toxic exposures.