Dietary Intake Is Associated with Biomarkers of Lead, Arsenic, and Cadmium in a Cohort of Mexican Adolescents

Abstract

Background

Lifestyle changes during adolescence can impact dietary habits and, subsequently, exposure to heavy metals.

Objectives

We aimed to evaluate the associations between food group intakes and metal exposures in a cohort of adolescents.

Methods

This study included 416 adolescents from Mexico City aged 10–18. Sociodemographic information at birth and repeated measurements of diet, anthropometry, and metal exposures were collected over 2 visits 3.5 ± 0.40 y apart (n = 514). Food groups (n = 31) were created based on the characteristics of 119 food frequency questionnaire items and metal dietary sources. Linear mixed-effect models were used to evaluate the associations between food group intake and exposure to blood lead, urinary arsenic, and urinary cadmium for the overall population (main models) and stratified by sex. Models were adjusted for age, sex (main models), maternal age, socioeconomic status, and specific gravity (only for urinary metals).

Results

Fruit intake in girls {2.63% [95% confidence interval (CI): 0.22%, 5.10%]}, and candy in boys [2.13% (95% CI: 0.40%, 3.88%)] and in the whole population [1.38% (95% CI: 0.16%, 2.61%)] were associated with higher blood lead levels. Additionally, leafy greens intake was associated with higher blood lead [10.75% (95% CI: –0.01%, 22.66%)]. Chicken intake in girls was associated with 5.95% (95% CI: 0.38%, 11.84%) higher urinary cadmium. Similarly, the intake of homemade sugar-sweetened beverages in girls [4.42% (35% CI: 0.13%, 8.89%)], and in the whole population [4.14% (95% CI: 1.42%, 6.94%)], was associated with higher urinary cadmium. Moreover, the intake of fish and seafood groups was positively associated with blood lead, urinary arsenic, and urinary cadmium.

Conclusions

We observed associations between food group intake and metal exposures in a group of Mexican adolescents using repeated measures of both outcomes and exposures. We also found that some of these associations varied by sex.

Introduction

Exposure to metals has been associated with the development of several adverse health outcomes in children. Exposure to lead has been linked to cognitive, behavioral, and academic problems in children [1,2]. Arsenic exposure has been associated with children’s neurodevelopment [3], higher risk of respiratory infections, asthma, cardiovascular outcomes, and cancers during childhood and adulthood [4]. Similarly, exposure to cadmium during childhood has been linked to altered renal biomarkers, bone resorption, and adverse cognitive and behavioral outcomes [2,5]. Diet represents the main source of heavy metal exposure in children [6,7]. However, most existing studies examining the relationship between diet and metal exposures have primarily focused on infants and young children, with the majority being cross-sectional studies [[8], [9], [10], [11], [12], [13]]. Similarly, the United States Food and Drug Administration (FDA)’s “Closer to Zero” initiative to reduce dietary exposure to mercury, lead, arsenic, and cadmium focuses primarily on prenatal or early childhood metal exposures [14]. Little is known about the role of diet as a source of metal exposures in adolescents.

In the adolescence period, nutrient requirements change, and adolescents become more independent and are able to make their own choices regarding food [15]. There are marked sex-specific changes in this group, as males eat larger food quantities, and are more likely to meet daily micronutrient requirements than girls [16]. Additionally, girls’ dietary patterns are often driven by external motivators, such as cultural factors (peer influence, managing appearance, and health status), whereas boys tend to make their food choices based on personal factors like food preferences and athletic performance goals [17]. The dietary changes that occur during adolescence not only influence long-term dietary habits [15], but could also affect their exposure to toxicants. Thus, diet during adolescence could represent an important target for preventing noncommunicable conditions in adulthood.

In this study, we evaluated the longitudinal associations of food group intakes and biomarkers of metal exposures in adolescents from Mexico City, and whether these associations varied by sex. We hypothesized that certain food groups would be associated with metal exposures measured in blood and urine according to current literature on metal exposure sources in the food supply chain, such as the 2018–2022 Total Diet Study report [18]. For example, we expected root vegetables and food items containing chocolate or chocolate flavoring to be associated with both lead and cadmium, leafy greens with all 3 evaluated metals (i.e., lead, arsenic, and cadmium), and vegetables other than leafy greens and root vegetables to be positively associated with arsenic and cadmium. Furthermore, we expected associations between nuts, whole and refined grains with cadmium, rice with arsenic, and fish and seafood groups with arsenic and lead [18]. The Mexican population is not exempt from dietary metal exposures. Traditional lead-glazed ceramics, commonly used in food serving and preparations, are considered a permanent and most important lead source for Mexican children [19,20]. However, other sources have been identified, for example, some commercial Mexican candies commonly consumed by young populations contain lead [21]. Also, drinking water has been reported as a source of arsenic nationwide [22]. Bovine liver and processed pork meat intakes have been associated with blood cadmium [23]. Specifically, in Mexican females, nonbottled water, eggs, fish, shellfish, root vegetables, and water-rich fruits and vegetables have been associated with total urinary arsenic [24], whereas nonstarchy vegetables, legumes, and processed meats have been associated with urinary cadmium [25].

Methods

Study sample

The current project derives from the ongoing Early Life Exposures in Mexico to Environmental Toxicants (ELEMENT) birth cohort, which recruited pregnant females from 1994 to 2006 who attended 3 prenatal clinics that serve a low- to middle-income population in Mexico City. The initial goals of this study were to evaluate 1) the detrimental effects of prenatal lead exposure and childhood neurodevelopment, and 2) the effect of calcium supplementation and lead-related outcomes. A detailed list of the females’ inclusion and exclusion criteria has been previously published [26,27]. Children were followed up from birth until adolescence at multiple time points, during which anthropometry, dietary information, and other visit-specific measurements or determinations were obtained (Figure 1). Briefly, between 2011 and 2012, a sample of 251 children (8–14 y old, peripuberty visit) were selected based on the availability of prenatal biospecimens. Finally, from 2015 to 2017, 558 individuals were re-recruited for 1 more visit (adolescence visit).

FIGURE 1.

Flowchart of the study population and analytical sample selection.

Complete blood and urine metal determinations were available for 649 observations in the peripuberty visit (n = 245) and the adolescence visit (n = 403). For this analysis, participants younger than 10 y of age (n = 124) were excluded based on the WHO’s definition of adolescence [28]. Finally, we removed observations with an energy intake >3 Z-score (≈ 4800 kcal) as they were considered implausible (n = 10). The current analytical sample included 416 participants who attended the peripuberty (n = 21), adolescence (n = 297), or both visits (n = 98) with complete exposure and outcome data, for a total of 514 observations (119 and 395 observations for the peripuberty and adolescent visits, respectively). The visits were on average 3.5 ± 0.40 y apart (Figure 1).

The Research, Ethics and Biosafety Committees of the National Institute of Public Health of Mexico, and the Institutional Review Board of the University of Michigan evaluated and approved this research protocol. Maternal informed consent and child assent were obtained for all participants.

Dietary assessment

During the peripuberty and adolescence visits, dietary information was collected using the 7-d recall food frequency questionnaire (FFQ) from the National Survey on Health and Nutrition from Mexico 2006, which included 119 items [29]. The participants were assisted by their caregiver with the recall or to provide more details when necessary. This instrument has been validated in Mexican adolescents using one 24-h recall, demonstrating a moderate validity for energy, macronutrient, and micronutrient intake, as well as good validity in the ranking of individuals according to nutrient intake [29].

The frequency of intake (weekly and daily) and portion sizes [standard portion sizes or measures of weight (g) and volume (mL)] for each food item were registered. This information was later modified to reflect the average weekly quantities consumed (mL or g/wk). Subsequently, 31 food groups were created based on items’ characteristics and sources of lead, cadmium, and arsenic from the 2018–2022 Total Diet Study report [18] (Supplemental Table 1). The intake units (mL or g/wk) of 26 of the 31 food groups were changed to reflect portion sizes per week in the Mexican population (e.g., 240 mL, ½ cup, etc., per week) [30]. The intake units of the remaining 5 food groups, Mexican food and 4 seafood groups, were changed to reflect the weight of standard items of 150 g (equivalent to, e.g., 2 enchiladas or 2 tacos [30]), and 28 g (corresponding to 25% of the United States FDA’s recommendation for children ≥ 11 y [31]), respectively. Finally, processed unsweetened beverages and high omega-3 nonpredatory seafood were not included in the final analysis as their intake was infrequent or had a low variation, respectively. This resulted in the final 29 food groups included in the statistical analysis.

Metal assessments

We included blood lead, total urinary arsenic, and urinary cadmium concentrations at both peripuberty and adolescence visits due to their importance in childhood development, as well as the current FDA’s food supply monitoring and initiative to reduce dietary exposure to lead, cadmium, arsenic, and mercury in young children [14]. Blood lead is a biomarker for recent exposure with a half-life of 1–2 mo, and its absorption after ingestion in children is 5 times greater than in adults [32]. Total urinary arsenic reflects recent exposure to organic and inorganic arsenic species and has a half-life of 2 d [33]. On the other hand, cadmium accumulates mainly in the kidneys, and urinary cadmium reflects long-term exposure with a half-life that spans from a few years to decades [34].

Blood lead determination

For the peripuberty visit, blood samples (2 mL) were collected and stored in trace-metal-free tubes using standardized protocols. Lead concentrations were determined in blood using graphite furnace atomic absorption spectroscopy (GFAAS) at American British Cowdray Hospital and validated by the Maternal and Child Health Bureau (MCHB) and the Wisconsin State Laboratory of Hygiene Cooperative Blood Lead Proficiency Testing Program (WSLH PBPTP) [35]. The quality control (QC) tests demonstrated adequate precision and accuracy for determinations (r = 0.99; mean difference of 0.17 mg/dL compared with MCHB and WSLH PBPTP blanks and spikes) [36,37]. All blood lead levels were above the limit of detection (LOD) of 0.4 ng/mL, and the precision of this instrument was within 1 μg/dL. For the adolescence visit, whole blood specimens and QC samples were analyzed on an inductively coupled plasma mass spectrometry (ICPMS), which used a 6-point calibration curve consisting of a calibration blank and standards 1–5, and a linearity check spanning from 0–100 μg/dL (Michigan Department of Health & Human Services laboratory). The specimens with correlation coefficients ≥ 0.997 were analyzed and subjected to QC runs at the end of the batch for instrument verification and drift. All blood lead levels were above the 0.284 μg/dL LOD.

Total urinary arsenic and urinary cadmium determinations

In both visits, urine samples were collected in sterile cups. The samples were then aliquoted, frozen, and stored at − 80°F, and shipped on dry ice for laboratory analyses. Urinary metals from the peripuberty visit were analyzed at McGill University. Urinary metals were measured using ICPMS (Varian 820-MS, Inc.) [38]. Accuracy and precision were measured using certified reference standards (Institut National de Santé Publique du Québec) with coefficients of variation ranging from 3% to 14%, and each batch run contained procedural blanks and replicate runs [39]. The metal analyses in urine from the adolescence visit were performed at the National Science Foundation International using an ICPMS assay method based on the Centers for Disease Control and Prevention (CDC) method 3018.3, with modifications for the expanded metals panel and the Thermo Scientific iCAP RQ instrument (serial number RQ0029). Arsenic and cadmium were analyzed using kinetic energy discrimination. All urinary arsenic concentrations were above the LOD of 0.3 μg/L, whereas 35.21% of the urinary cadmium levels were below the LOD of 0.06 μg/L. The original <LOD urinary cadmium values provided by the instrument were used in our analysis. Urinary specific gravity (SG) was measured using a handheld digital refractometer (Atago Co., Ltd.) [40].

Covariates

Upon enrollment, mothers reported their age in y, education level (elementary, middle, high school, and undergraduate/graduate school), smoking habits during pregnancy, and household socioeconomic status (SES) via interviewer-administered questionnaires. SES was estimated using a validated scale consisting of 13 questions on housing quality, health infrastructure, services, communication and entertainment technology, head of household education, and household expenses [41,42]. The questionnaire defined 6 SES categories using hierarchical trees (A/B, C+, C, D+, D, E, where A/B was the highest). Adolescents’ sex information was obtained from the historical records of the cohort. The energy for each food item reported in the FFQ was calculated using a nutritional composition database of foods compiled by the National Institute of Public Health of Mexico [43]. This information was later used to estimate the daily total energy intake (kcal/d) based on the intake frequency and portion sizes of the food items reported by each individual.

Age, sex, maternal age, and low SES were considered confounders due to their association with dietary intake and eating habits [16,44], as well as metal exposures [7,45,46]. Similarly, we considered daily energy intake as a confounder due to its association with physical activity, nutrient intake, and metabolic efficiency [47]. The year of visit (2011–2012 or 2015–2016) was included in the models to control for the different blood lead determination methods used in the peripuberty (2011–2012) and adolescence visits (2015–2017) (GFAAS, and ICPMS, respectively), as well as the contrasting proportion of participants who attended either visit or both (71.4%, 5.0% and 23.6%, respectively). SES was preferred over maternal education for the analyses, as SES was a better predictor of blood lead. We decided not to include BMI (in kg/m²) in our models, as the adjustment for this variable could have led to collider bias in the association between food group intakes and metal exposures. In other words, BMI could be a common effect of both our exposure and outcome [48]. Smoking during pregnancy was not considered a confounder in our analyses, as only 1.95% (n = 10) of the females reported having smoked. Similarly, smoking was not included in the main analysis because this variable was only available for the adolescent visit (n = 392), where only 16.58% of adolescents responded yes to having ever smoked (n = 65). A sensitivity analysis was performed to evaluate the effect of smoking on cadmium models for the adolescent visit.

Statistical analysis

In the analytical sample, missing specific gravity values (3.3% of missingness) were replaced by the average specific gravity in the whole population (mean = 1.01954). Missing observations for the SES variable (12.26% of the analytical sample) were replaced using single imputation with the MICE R package (version 3.16.0). The SES variable was collapsed into low and high SES (D to E compared with A/B to D+). Blood lead, urinary arsenic, and urinary cadmium concentrations were log-transformed to normalize their right-skewed distributions.

Linear mixed-effects models (LMM) were used to evaluate the associations between the dietary exposures and concentrations of each metal. The LMM models were performed using the nlme R package (version 3.1-163). Random intercepts were included in all models to account for temporal dependence, whereas interaction terms between the food group intake and age were considered to improve the models’ goodness of fit. The inclusion of random slopes was not supported by the data, and these models failed to converge. All models were adjusted for age, maternal age, sex, visit year, total energy intake (kcal/d), low SES, and specific gravity (only for urinary arsenic and cadmium models). Sex-stratified LMM were also performed to account for potential sex differences in dietary patterns [16,17], heavy metal absorption, accumulation, and biotransformation [49]. Sex-stratified models were adjusted for the same variables except for sex. The main analysis comprised 29 independent crude and adjusted models for each of the 3 metals, for a total of 174 LMM. The sex-stratified analysis consisted of 348 LMM. To account for the possibility of false positives associated with multiple comparisons in both crude and adjusted models, false discovery rate (FDR) corrections [50] were performed at the 5% FDR level for 29 comparisons for each of the 3 metals evaluated. The FDR can be defined as the expected proportion of false positives as or more extreme than the observed one [50].

We performed several sensitivity analyses. First, although we had originally considered drinks containing natural fruit as part of the homemade sugar-sweetened (juices and fruit cooler drinks with added sugar) or homemade unsweetened beverages (fruit juices without added sugar) (Supplemental Table 1), due to fruits’ potential arsenic content [24,51], alternative food groups for items containing fruit were created regardless of their presentation (fruit + natural fruit juice and homemade drinks with natural fruit juice). Also, we compared our main urinary arsenic and cadmium models adjusted for the imputed SG variable against models using the unimputed SG variable (514 compared with 497 observations). Similarly, we evaluated the impact of urinary cadmium concentrations < LOD by replacing them with the mean of the values < LOD (0.043 μg/L; min, max: 0.10, 0.06). The effect of potential residual confounding by other arsenic and cadmium sources was evaluated by adjusting these models for drinking water sources (n = 514; bottled water, public or private source) and smoking habits during the adolescence visit (n = 392), respectively. Furthermore, a cross-sectional analysis was performed only considering the adolescence visit (n = 395), as only 23.6% of the participants had information for both peripuberty and adolescence visits, and given the different lead determination methods used in both visits (GFAAS and ICPMS, respectively). Girls’ models were additionally adjusted for menarche (yes/no) at the time of interview as a proxy variable for menstruation, as blood lead levels seem to increase during this time [52]. For this sensitivity analysis, 0.38% of the menarche observations were imputed. Finally, to differentiate between organic and inorganic arsenic, urinary arsenic models were further stratified by fish and seafood intake (<28 g compared with ≥28 g) as their intake is an important source of organic arsenic [53].

All coefficients and 95% confidence intervals (CI) are presented as the percentage change in each of the metal’s concentration by exponentiating, subtracting 1, and multiplying by 100. All the analyses were performed using R version 4.2.2.

Results

Compared with the children who attended the 2 previous study visits (childhood visits, Figure 1), the participants part of the peripuberty and adolescent visits had lower BMIs 2.22 ± 0.72 y before the peripuberty visit [median (IQR): 17.31 kg/m² (15.84, 20.01) compared with 18.79 kg/m² (16.65, 22.01; P < 0.001)], and their mothers where on average slightly older at their birth [median (IQR): 25.00 y (22.00, 29.00) compared with 26.00 y (23.00, 30.00), respectively, P = 0.004]. Most of the included participants attended only the adolescence visit (71.4%), and only 5.0% attended the peripuberty visit. The remaining participants had complete information for both visits (23.6%). Of the participants included in the analysis (n = 416), 51.7% were females. The participants’ characteristics at baseline are presented in Table 1. The population had median blood lead levels (2.40 μg/dL) under the reference value from the CDC of 3.5 μg/dL [54]. Similarly, the median total urinary arsenic concentrations (11.83 μg/L) were considerably below the 50 μg/L threshold associated with minor health risks [55]. Finally, the median urinary cadmium observed in this population (0.09 μg/L) was below the geometric mean for individuals 6 y and older (0.19 μg/L) [56]. In our population, compared with girls, boys had higher blood lead [median (IQR): 2.20 μg/dL (1.50, 3.15) compared with 2.50 μg/dL (1.80, 3.70), respectively; P = 0.010] and higher urinary arsenic concentrations [median (IQR): 11.17 μg/L (7.37, 16.20) compared with 12.35 μg/L (8.43, 18.23), respectively; P = 0.056]. Also, compared with girls, boys had higher energy intakes and consumed more processed sugar-sweetened beverages (SSBs), unsweetened dairy, eggs, red meat, processed meat, rice, refined grains, corn, soups and creams, and Mexican and fast food. Girls, on the other hand, had higher candy, fish, and seafood intakes. No sex differences were observed for age, maternal age, and household SES.

TABLE 1.

Baseline population characteristics and dietary intake.

	Overall (n = 416)	Boys (n = 201)	Girls (n = 215)	P1
	Median (IQR); n (%)	Median (IQR); n (%)	Median (IQR); n (%)
Age (y)	12.60 (11.60, 13.72)	12.70 (11.60, 13.80)	12.40 (11.60, 13.60)	0.342
Year of visit				0.693
2011	38 (9.13%)	15 (7.46%)	23 (10.70%)
2012	81 (19.47%)	41 (20.40%)	40 (18.60%)
2015	209 (50.24%)	103 (51.24%)	106 (49.30%)
2016	88 (21.15%)	42 (20.90%)	46 (21.40%)
Energy intake (kcal)	2184.41 (1740.53, 2782.75)	2420.94 (1899.85, 3086.38)	1988.96 (1641.11, 2489.39)	<0.001
Low SES at pregnancy	225 (54.09%)	105 (52.24%)	120 (55.81%)	0.465
Maternal age at birth	26.0 (23.00, 30.00)	26.0 (22.00, 30.00)	26.0 (23.00, 30.00)	0.308
Metal exposure concentrations
Blood lead (μg/dL)	2.40 (1.70, 3.50)	2.50 (1.80, 3.70)	2.20 (1.50, 3.15)	0.010
Urinary arsenic (μg/L)	11.83 (7.85, 17.24)	12.35 (8.43, 18.23)	11.17 (7.37, 16.20)	0.056
Urinary cadmium (μg/L)	0.09 (0.04, 0.14)	0.09 (0.04, 0.14)	0.08 (0.04, 0.15)	0.782
Water and seafood intake
Drinking water source at home				0.454
Water jug/water delivery service	296 (71.15%)	141 (70.15%)	155 (72.09%)
Public	82 (19.71%)	44 (21.89%)	38 (17.67%)
Private	38 (9.13%)	16 (7.96%)	22 (10.23%)
Fish/seafood consumers (≥28 g/wk)	254 (61.06%)	109 (54.23%)	145 (67.44%)	0.006
High omega 3 predatory seafood consumers (≥ 28 g/wk)	129 (31.01%)	62 (30.85%)	67 (31.16%)	0.944
Low omega 3 predatory seafood consumers (≥ 28 g/wk)	186 (44.71%)	71 (35.32%)	115 (53.49%)	<0.001
Low omega 3 nonpredatory seafood consumers (≥ 28 g/wk)	43 (10.34%)	19 (9.45%)	24 (11.16%)	0.567
Weekly food group intake
Processed SSB (240 mL)	3.92 (1.33, 7.50)	5.33 (2.00, 9.00)	3.33 (1.00, 7.42)	0.062
Homemade SSB (240 mL)	7.50 (3.00, 16.50)	8.00 (3.00, 18.50)	7.50 (3.00, 14.00)	0.221
Homemade unsweetened drinks (240 mL)	17.50 (7.00, 31.50)	17.50 (7.00, 34.50)	17.50 (7.00, 28.62)	0.082
Unsweetened dairy (240 mL)	13.36 (6.83, 20.07)	18.58 (8.01, 20.47)	10.96 (5.17, 19.71)	0.001
Sweetened dairy (240 mL)	1.43 (0.33, 2.35)	1.48 (0.28, 2.51)	1.31 (0.42, 2.28)	0.959
Leafy vegetables (90 g)	1.94 (0.67, 4.00)	1.67 (0.33, 4.00)	2.00 (0.67, 3.97)	0.356
Root vegetables (65 g)	2.69 (1.08, 7.12)	2.69 (1.08, 6.92)	2.69 (1.08, 7.12)	0.720
Other veggies (90 g)	5.57 (2.71, 10.32)	5.78 (3.03, 10.32)	5.16 (2.54, 10.28)	0.566
Fruit (90 g)	16.38 (8.22, 28.72)	16.50 (8.07, 28.64)	16.26 (8.63, 28.77)	0.838
Chicken (30 g)	9.00 (3.00, 11.33)	9.00 (3.00, 13.67)	9.00 (3.00, 11.00)	0.055
Eggs (50 g)	3.30 (1.10, 6.60)	3.30 (2.20, 6.60)	2.34 (1.10, 6.60)	0.022
Red meat (30 g)	5.50 (2.57, 11.00)	5.50 (3.67, 13.13)	3.85 (1.83, 8.00)	<0.001
Processed meat (30 g)	16.00 (8.92, 27.83)	19.00 (13.00, 30.33)	15.00 (6.33, 24.83)	<0.001
Rice (1/4 cup or 50 g)	6.00 (2.00, 6.00)	6.00 (2.00, 6.00)	5.50 (2.00, 6.00)	<0.001
Refined grains (30 g)	10.67 (5.50, 18.81)	11.83 (6.00, 19.50)	10.00 (5.33, 17.46)	0.127
Corn (30 g)	17.50 (10.00, 23.33)	17.50 (11.67, 30.50)	14.17 (9.00, 22.50)	<0.001
Whole grains (30 g)	0.00 (0.00, 0.00)	0.00 (0.00, 0.00)	0.00 (0.00, 1.17)	0.403
Potatoes (60 g)	0.67 (0.00, 2.00)	0.67 (0.00, 2.00)	0.67 (0.00, 1.33)	0.105
Legume (1/2 or 90g)	1.39 (0.56, 2.22)	1.67 (0.56, 3.33)	1.11 (0.56, 2.22)	0.193
Nuts (15 g)	0.00 (0.00, 2.33)	0.00 (0.00, 2.33)	0.00 (0.00, 2.33)	0.639
Soups and creams (240 mL)	1.25 (0.63, 2.25)	1.42 (0.82, 2.45)	1.13 (0.62, 2.25)	0.009
Mexican food (150 g)	1.81 (0.67, 3.68)	2.00 (0.67, 4.00)	1.33 (0.67, 3.15)	0.129
Fast food (100 g)	1.05 (0.35, 1.97)	1.05 (0.35, 2.19)	0.92 (0.35, 1.93)	0.104
Candy (10 g)	6.00 (0.00, 21.00)	6.00 (0.00, 18.00)	9.00 (1.00, 22.75)	0.010
Pastries and desserts (20 g)	20.25 (10.50, 36.18)	21.00 (10.50, 36.00)	20.10 (9.97, 36.40)	0.937
Spreads (5 g)	2.00 (0.00, 6.00)	2.00 (0.00, 6.00)	2.00 (0.00, 6.00)	0.721
High omega 3 predatory seafood (28 g)	0.005 (0.00, 1.59)	0.008 (0.00, 1.59)	0.005 (0.00, 1.59)	0.854
Low omega 3 predatory seafood (28 g)	0.04 (0.008, 2.83)	0.02 (0.005, 1.43)	1.41 (0.02, 2.86)	<0.001
Low omega 3 nonpredatory seafood (28 g)	0.00 (0.00, 0.008)	0.00 (0.00, 0.008)	0.00 (0.00, 0.008)	0.506

Abbreviation: SSBs, sugar-sweetened beverages.

¹Wilcoxon rank sum test; Pearson's Chi-squared test.

After adjusting for confounders, a 10 g higher intake of candy was associated with a 1.38% (95% CI: 0.16%, 2.61%) higher blood lead (Table 2 and Supplemental Figure 1). Similarly, the intake of leafy vegetables was associated with 10.75% (95% CI: –0.01%, 22.66%; P = 0.052) higher blood lead levels, where 41.1% of false positives were expected (Table 2 and Supplemental Figure 1). Higher intake of low omega-3 predatory seafood and low omega-3 nonpredatory seafood by 28 g were associated with 6.52% (95% CI: 0.79%, 12.58%) and 54.93% (95% CI: –0.59%, 141.46%; P = 0.056) higher urinary arsenic, respectively. On the other hand, higher intakes of ½ cup of root vegetables and ¼ cup of rice were associated with a 6.26% (95% CI: –11.81%, –0.35%) and 7.86% (95% CI: –14.95%, –0.18%) lower in urinary arsenic, respectively. At an FDR q = 0.05, these associations between food groups and urinary arsenic had an expected 40.5% of false positives (Table 3 and Supplemental Figure 2). Higher homemade SSBs (240 mL) intake was associated with higher urinary cadmium by 4.14% (95% CI: 1.42%, 6.94%) (Table 4 and Supplemental Figure 3), where 9.80% of false positives were expected.

TABLE 2.

Associations between log-transformed blood lead concentrations and food group intakes.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.28	–0.36, 0.93	0.388	0.626	3.68	–0.64, 8.19	0.098	0.411
Homemade SSB (240 mL)	0.30	–0.15, 0.76	0.194	0.613	–0.21	–3.34, 3.01	0.894	0.961
Homemade unsweetened drinks (240 mL)	–0.19	–0.50, 0.13	0.244	0.626	0.18	–1.70, 2.09	0.852	0.961
Unsweetened dairy (240 mL)	–0.26	–0.82, 0.30	0.359	0.626	–1.94	–5.63, 1.90	0.319	0.645
Sweetened dairy (240 mL)	–2.03	–4.54, 0.54	0.120	0.531	1.17	–13.48, 18.31	0.884	0.961
High omega-3 predatory seafood (28 g)	0.38	–1.95, 2.77	0.747	0.833	13.55	–1.84, 31.36	0.090	0.411
Low omega-3 predatory seafood (28 g)	–0.28	–0.89, 0.34	0.377	0.626	1.60	–4.37, 7.95	0.607	0.838
Low omega-3 nonpredatory seafood (28 g)	–4.97	–11.12, 1.60	0.134	0.531	12.54	–27.66, 75.08	0.601	0.838
Leafy vegetables (90 g)	–0.28	–1.88, 1.35	0.734	0.833	10.75	–0.01, 22.66	0.052	0.411
Root vegetables (65 g)	0.69	–0.24, 1.63	0.147	0.531	1.76	–4.37, 8.28	0.583	0.838
Potatoes (60 g)	1.29	–1.29, 3.94	0.327	0.626	–11.20	–25.79, 6.26	0.197	0.567
Other veggies (90 g)	0.22	–0.47, 0.92	0.536	0.740	3.73	–0.55, 8.20	0.091	0.411
Fruit (90 g)	0.08	–0.14, 0.31	0.462	0.705	0.65	–0.87, 2.20	0.406	0.736
Chicken (30 g)	0.81	0.26, 1.36	0.004	0.041	2.12	–1.48, 5.84	0.254	0.567
Eggs (50 g)	0.02	–1.48, 1.54	0.982	0.982	3.83	–5.71, 14.34	0.445	0.760
Soups and creams (240 mL)	1.92	–1.09, 5.03	0.211	0.613	12.44	–7.69, 36.95	0.246	0.567
Refined grains (30 g)	–0.21	–0.67, 0.26	0.373	0.626	0.28	–2.76, 3.42	0.857	0.961
Whole grains (30 g)	–0.45	–1.72, 0.84	0.488	0.707	8.13	–0.79, 17.84	0.078	0.411
Corn (30 g)	0.70	0.32, 1.08	0.0004	0.011	–0.11	–2.57, 2.41	0.930	0.963
Rice (1/4 cup or 50 g)	2.10	0.85, 3.37	0.001	0.018	7.33	–1.26, 16.68	0.099	0.411
Red meat (30 g)	–0.06	–0.52, 0.41	0.811	0.840	1.05	–2.27, 4.48	0.542	0.838
Processed meat (30 g)	0.19	–0.19, 0.57	0.319	0.626	–1.81	–4.08, 0.52	0.129	0.466
Legume (90g)	2.33	–0.13, 4.86	0.063	0.431	8.79	–8.24, 28.98	0.334	0.645
Nuts (15 g)	–0.65	–1.83, 0.54	0.279	0.626	–0.22	–10.23, 10.90	0.967	0.967
Mexican food (150 g)	0.40	–1.31, 2.15	0.643	0.810	7.69	–4.24, 21.09	0.218	0.567
Fast food (100 g)	0.38	–2.39, 3.24	0.787	0.840	4.57	–13.28, 26.10	0.640	0.844
Candy (10 g)	0.19	–0.02, 0.39	0.074	0.431	1.38	0.16, 2.61	0.029	0.411
Pastries and desserts (20 g)	–0.05	–0.29, 0.19	0.690	0.833	–1.09	–2.64, 0.48	0.173	0.557
Spreads (5 g)	–0.21	–1.02, 0.61	0.612	0.807	0.40	–4.81, 5.90	0.882	0.961

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

TABLE 3.

Associations between log-transformed urinary arsenic concentrations and food group intakes.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.01	–0.63, 0.64	0.982	0.991	2.56	–1.82, 7.14	0.258	0.681
Homemade SSB (240 mL)	–0.10	–0.54, 0.34	0.655	0.991	–1.21	–4.16, 1.83	0.432	0.746
Homemade unsweetened drinks (240 mL)	–0.06	–0.36, 0.25	0.718	0.991	–0.57	–2.43, 1.34	0.559	0.771
Unsweetened dairy (240 mL)	0.003	–0.54, 0.55	0.991	0.991	2.10	–1.75, 6.10	0.291	0.704
Sweetened dairy (240 mL)	–0.49	–2.93, 2.01	0.693	0.991	–1.23	–16.34, 16.62	0.884	0.950
High omega-3 predatory seafood (28 g)	5.97	3.52, 8.48	<0.00001	0.0001	–11.54	–23.07, 1.73	0.088	0.512
Low omega-3 predatory seafood (28 g)	0.70	0.12, 1.28	0.019	0.134	6.52	0.79, 12.58	0.028	0.405
Low omega-3 nonpredatory seafood (28 g)	16.51	9.05, 24.48	0.00001	0.0001	54.93	–0.59, 141.46	0.056	0.405
Leafy vegetables (90 g)	0.64	–0.91, 2.22	0.418	0.772	4.94	–5.27, 16.25	0.358	0.741
Root vegetables (65 g)	0.73	–0.21, 1.68	0.127	0.409	–6.26	–11.81, –0.35	0.041	0.405
Potatoes (60 g)	0.13	–2.29, 2.60	0.919	0.991	–4.02	–19.24, 14.07	0.642	0.846
Other veggies (90 g)	0.27	–0.39, 0.93	0.426	0.772	1.28	–2.75, 5.49	0.539	0.771
Fruit (90 g)	0.02	–0.20, 0.24	0.838	0.991	–0.63	–2.15, 0.92	0.425	0.746
Chicken (30 g)	0.02	–0.51, 0.56	0.942	0.991	2.12	–1.46, 5.82	0.252	0.681
Eggs (50 g)	1.11	–0.72, 2.97	0.235	0.510	–5.36	–13.82, 3.93	0.251	0.681
Soups and creams (240 mL)	–2.44	–5.27, 0.48	0.100	0.379	–11.14	–26.39, 7.27	0.221	0.681
Refined grains (30 g)	0.31	–0.15, 0.77	0.188	0.496	0.08	–2.97, 3.23	0.958	0.958
Whole grains (30 g)	0.12	–1.12, 1.36	0.855	0.991	–1.50	–9.73, 7.48	0.735	0.926
Corn (30 g)	0.08	–0.29, 0.44	0.683	0.991	–0.08	–2.48, 2.37	0.947	0.958
Rice (1/4 cup or 50 g)	–0.07	–1.26, 1.13	0.905	0.991	–7.86	–14.95, –0.18	0.048	0.405
Red meat (30 g)	–0.43	–0.86, 0.01	0.057	0.333	–2.36	–5.69, 1.09	0.181	0.681
Processed meat (30 g)	0.23	–0.13, 0.60	0.211	0.510	1.50	–0.84, 3.90	0.213	0.681
Legume (90g)	–1.92	–4.18, 0.40	0.104	0.379	–2.37	–17.20, 15.12	0.776	0.937
Nuts (15 g)	0.29	–0.93, 1.53	0.638	0.991	3.69	–6.69, 15.23	0.502	0.771
Mexican food (150 g)	–2.29	–3.90, –0.66	0.007	0.065	4.51	–6.47, 16.77	0.437	0.746
Fast food (100 g)	–1.83	–4.48, 0.89	0.183	0.496	–8.38	–23.81, 10.18	0.354	0.741
Candy (10 g)	–0.01	–0.20, 0.19	0.942	0.991	0.39	–0.86, 1.66	0.540	0.771
Pastries and desserts (20 g)	0.14	–0.10, 0.38	0.246	0.510	–0.12	–1.69, 1.47	0.881	0.950
Spreads (5 g)	0.71	–0.13, 1.54	0.097	0.379	–0.68	–6.00, 4.95	0.810	0.939

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

TABLE 4.

Associations between log-transformed urinary cadmium concentrations and food group intakes.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	–0.21	–0.83, 0.42	0.518	0.693	0.71	–3.06, 4.63	0.717	0.927
Homemade SSB (240 mL)	–0.17	–0.61, 0.27	0.436	0.668	4.14	1.42, 6.94	0.003	0.098
Homemade unsweetened drinks (240 mL)	–0.15	–0.45, 0.15	0.325	0.589	0.25	–1.40, 1.93	0.770	0.927
Unsweetened dairy (240 mL)	0.05	–0.49, 0.59	0.860	0.922	–0.26	–3.58, 3.16	0.878	0.927
Sweetened dairy (240 mL)	3.00	0.51, 5.56	0.019	0.344	–3.67	–16.57, 11.22	0.611	0.927
High omega-3 predatory seafood (28 g)	1.82	–0.57, 4.27	0.136	0.375	1.55	–10.52, 15.24	0.812	0.927
Low omega-3 predatory seafood (28 g)	–0.43	–1.00, 0.15	0.142	0.375	–0.68	–5.49, 4.36	0.786	0.927
Low omega-3 nonpredatory seafood (28 g)	6.63	–0.27, 14.00	0.061	0.344	27.26	–14.41, 89.23	0.236	0.741
Leafy vegetables (90 g)	1.66	0.11, 3.24	0.037	0.344	0.60	–8.02, 10.03	0.896	0.927
Root vegetables (65 g)	0.62	–0.31, 1.56	0.189	0.422	–1.90	–7.07, 3.57	0.490	0.927
Potatoes (60 g)	0.50	–2.88, 1.95	0.685	0.828	11.32	–4.37, 29.58	0.169	0.622
Other veggies (90 g)	0.18	–0.47, 0.84	0.583	0.736	0.83	–2.73, 4.51	0.654	0.927
Fruit (90 g)	0.17	–0.04, 0.39	0.117	0.375	–0.70	–2.03, 0.65	0.310	0.749
Chicken (30 g)	–0.17	–0.70, 0.36	0.525	0.693	2.74	–0.43, 0.60	0.094	0.622
Eggs (50 g)	–0.89	–2.31, 0.56	0.227	0.471	–6.29	–13.66, 1.71	0.123	0.622
Soups and creams (240 mL)	–2.39	–5.20, 0.51	0.106	0.375	3.63	–12.23, 22.35	0.674	0.927
Refined grains (30 g)	0.42	–0.04, 0.88	0.071	0.344	–0.86	–3.50, 1.86	0.534	0.927
Whole grains (30 g)	0.21	–1.02, 1.45	0.738	0.856	–6.42	–13.29, 1.00	0.091	0.622
Corn (30 g)	0.04	–0.32, 0.40	0.825	0.921	–0.14	–2.25, 2.02	0.898	0.927
Rice (1/4 cup or 50 g)	0.05	–1.14, 1.25	0.938	0.938	–4.85	–11.35, 2.13	0.172	0.622
Red meat (30 g)	0.03	–0.41, 0.47	0.890	0.922	–1.66	–4.60, 1.37	0.281	0.741
Processed meat (30 g)	0.36	–0.002, 0.73	0.052	0.344	0.26	–1.78, 2.33	0.808	0.927
Legume (90g)	–0.92	–3.20, 1.41	0.432	0.668	0.68	–12.93, 16.42	0.927	0.927
Nuts (15 g)	–0.48	–1.69, 0.74	0.437	0.668	–7.03	–15.26, 1.99	0.126	0.622
Mexican food (150 g)	–0.62	–2.25, 1.04	0.462	0.671	–0.70	–10.03, 9.59	0.889	0.927
Fast food (100 g)	1.49	–1.23, 4.29	0.285	0.551	0.90	–14.18, 18.62	0.914	0.927
Candy (10 g)	0.13	–0.06, 0.33	0.179	0.422	0.63	–0.47, 1.74	0.265	0.741
Pastries and desserts (20 g)	0.27	0.03, 0.50	0.026	0.344	1.09	–0.31, 2.50	0.130	0.622
Spreads (5 g)	0.62	–0.21, 1.46	0.141	0.375	–1.89	–6.48, 2.92	0.436	0.927

Abbreviations: FDR, false discovery rate; CI, confidence interval; SSBs, sugar-sweetened beverages.

In sex-stratified analyses, in boys, higher intakes of candy by 10 g were associated with 2.13% (95% CI: 0.40%, 3.88%) higher blood lead. Also, in boys, a 240 mL higher intake of sweetened dairy was associated with a 24.95% (95% CI: –40.55%, –5.24%) lower blood lead. For a q = 0.05, these associations between food groups and blood lead in boys had an expected 19.3% of false positives (Table 5 and Supplemental Figure 4). In girls, higher intakes of sweetened dairy (240 mL), and high omega-3 predatory seafood (28 g) were associated with increased blood lead levels by 29.62% (95% CI: 1.80%, 65.05%), and 27.60% (95% CI: 3.11%, 57.92%), respectively. Also, in girls, increased intake of fruits (2.63%; 95% CI: 0.22%, 5.10%) and vegetables other than leafy and root varieties (7.74%; 95% CI: 1.76%, 14.06%) were associated with higher blood lead concentrations (Table 6 and Supplemental Figure 4). For the mentioned associations between food groups and blood lead for girls, 28.6% of false positives were expected.

TABLE 5.

Associations between log-transformed blood lead and food group intakes in boys.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.26	–0.62, 1.14	0.563	0.831	5.59	–0.71, 12.29	0.088	0.604
Homemade SSB (240 mL)	0.26	–0.33, 0.85	0.376	0.810	–3.09	–7.27, 1.27	0.166	0.604
Homemade unsweetened drinks (240 mL)	–0.20	–0.61, 0.21	0.328	0.810	–0.72	–3.12, 1.74	0.563	0.878
Unsweetened dairy (240 mL)	–0.37	–1.14, 0.40	0.337	0.810	–1.54	–6.70, 3.90	0.571	0.878
Sweetened dairy (240 mL)	–2.81	–6.13, 0.63	0.107	0.494	–24.95	–40.55, –5.24	0.019	0.193
High omega 3 predatory seafood (28 g)	0.95	–2.73, 4.76	0.613	0.836	2.40	–16.66, 25.82	0.821	0.923
Low omega-3 predatory seafood (28 g)	–0.28	–0.95, 0.40	0.411	0.810	–0.33	–7.96, 7.94	0.936	0.968
Low omega 3 nonpredatory seafood (28 g)	–3.47	–12.95, 7.04	0.497	0.810	50.45	–18.40, 177.41	0.194	0.604
Leafy vegetables (90 g)	–0.20	–2.60, 2.25	0.867	0.914	3.82	–10.58, 20.55	0.622	0.888
Root vegetables (65 g)	–0.08	–1.35, 1.21	0.902	0.914	–5.57	–13.73, 3.36	0.217	0.604
Potatoes (60 g)	0.46	–3.41, 4.48	0.817	0.914	–2.91	–26.39, 28.06	0.834	0.923
Other veggies (90 g)	–0.13	–1.08, 0.82	0.784	0.914	–0.59	–6.99, 6.26	0.862	0.923
Fruit (90 g)	0.18	–0.10, 0.47	0.203	0.610	–0.82	–2.75, 1.14	0.410	0.794
Chicken (30 g)	0.98	0.32, 1.65	0.005	0.072	1.57	–2.89, 6.23	0.497	0.877
Eggs (50 g)	–0.39	–2.36, 1.62	0.699	0.912	7.83	–5.42, 22.95	0.262	0.604
Soups and creams (240 mL)	2.60	–1.35, 6.71	0.196	0.609	17.92	–8.79, 52.45	0.211	0.604
Refined grains (30 g)	0.23	–0.40, 0.86	0.468	0.810	0.73	–3.35, 4.98	0.731	0.926
Whole grains (30 g)	–1.07	–2.45, 0.33	0.132	0.494	5.97	–3.12, 15.91	0.208	0.604
Corn (30 g)	0.89	0.43, 1.36	0.000	0.009	–0.06	–3.19, 3.17	0.971	0.971
Rice (1/4 cup or 50 g)	1.94	0.31, 3.59	0.021	0.211	6.55	–4.54, 18.93	0.260	0.604
Red meat (30 g)	–0.04	–0.57, 0.50	0.894	0.914	2.42	–1.42, 6.41	0.223	0.604
Processed meat (30 g)	0.13	–0.33, 0.59	0.582	0.831	–2.31	–5.08, 0.54	0.116	0.604
Legume (90 g)	2.99	–0.58, 6.69	0.101	0.494	10.35	–13.29, 40.45	0.424	0.794
Nuts (15 g)	–0.57	–1.85, 0.72	0.377	0.809	1.86	–10.90, 16.44	0.787	0.923
Mexican food (150 g)	0.83	–1.54, 3.25	0.491	0.810	9.59	–7.67, 30.08	0.296	0.635
Fast food (100 g)	–0.38	–3.72, 3.08	0.825	0.914	–3.53	–23.99, 22.45	0.767	0.923
Candy (10 g)	0.26	–0.06, 0.58	0.115	0.494	2.13	0.40, 3.88	0.019	0.193
Pastries and desserts (20 g)	0.02	–0.30, 0.34	0.914	0.914	–0.37	–2.37, 1.67	0.720	0.923
Spreads (5 g)	–0.40	–1.60, 0.82	0.513	0.810	2.23	–5.56, 10.65	0.585	0.878

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

TABLE 6.

Associations between log-transformed blood lead and food group intakes in girls.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.16	–0.80, 1.13	0.737	0.854	3.77	–3.07, 11.08	0.289	0.645
Homemade SSB (240 mL)	0.19	–0.54, 0.93	0.609	0.844	2.70	–2.15, 7.79	0.283	0.645
Homemade unsweetened drinks (240 mL)	–0.30	–0.81, 0.20	0.234	0.843	1.56	–1.48, 4.70	0.320	0.651
Unsweetened dairy (240 mL)	–0.31	–1.14, 0.54	0.471	0.844	–1.07	–6.67, 4.87	0.717	0.928
Sweetened dairy (240 mL)	–1.60	–5.31, 2.27	0.407	0.843	29.62	1.80, 65.05	0.039	0.286
High omega-3 predatory seafood (28 g)	0.10	-3.03, 3.33	0.950	0.950	27.60	3.11, 57.92	0.029	0.286
Low omega-3 predatory seafood (28 g)	–0.33	–1.85, 1.21	0.668	0.844	3.97	–6.87, 16.06	0.488	0.744
Low omega-3 nonpredatory seafood (28 g)	–5.64	–13.74, 3.22	0.201	0.843	–9.99	–52.41, 70.25	0.746	0.928
Leafy vegetables (90 g)	–0.61	–2.78, 1.61	0.581	0.844	13.96	–0.96, 31.13	0.072	0.350
Root vegetables (65 g)	1.39	0.03, 2.78	0.047	0.547	6.65	–2.23, 16.33	0.150	0.484
Potatoes (60 g)	1.58	–1.89, 5.17	0.372	0.843	–17.20	–35.62, 6.48	0.145	0.484
Other veggies (90 g)	0.48	–0.53, 1.50	0.350	0.843	7.74	1.76, 14.06	0.013	0.286
Fruit (90 g)	–0.08	–0.44, 0.28	0.655	0.844	2.63	0.22, 5.10	0.036	0.286
Chicken (30 g)	0.25	–0.70, 1.22	0.599	0.844	5.43	–0.87, 12.12	0.097	0.401
Eggs (50 g)	–0.32	–2.64, 2.05	0.783	0.874	2.15	–13.10, 20.06	0.796	0.928
Soups and creams (240 mL)	0.32	–4.27, 5.13	0.892	0.931	8.89	–20.75, 49.61	0.599	0.868
Refined grains (30 g)	–0.69	–1.37, –0.01	0.048	0.547	0.68	–4.04, 5.63	0.783	0.928
Whole grains (30 g)	1.28	–1.57, 4.20	0.378	0.843	16.51	–7.56, 46.84	0.198	0.523
Corn (30 g)	0.14	–0.52, 0.82	0.670	0.844	–0.44	–4.43, 3.72	0.832	0.928
Rice (1/4 cup or 50 g)	1.92	–0.05, 3.93	0.057	0.547	14.49	–0.97, 32.36	0.072	0.350
Red meat (30 g)	–0.71	–1.71, 0.31	0.167	0.844	–0.40	–7.32, 7.03	0.912	0.979
Processed meat (30 g)	0.15	–0.51, 0.81	0.655	0.844	0.03	–4.07, 4.29	0.991	0.991
Legume (90 g)	1.51	–1.89, 5.02	0.383	0.843	10.01	–15.32, 42.93	0.475	0.745
Nuts (15 g)	–1.19	–3.74, 1.43	0.363	0.843	–0.16	–16.27, 19.05	0.986	0.991
Mexican food (150 g)	–0.57	–3.05, 1.98	0.655	0.844	8.01	–8.37, 27.32	0.359	0.651
Fast food (100 g)	0.86	–3.86, 5.82	0.722	0.854	14.71	–16.68, 57.92	0.401	0.683
Candy (10 g)	0.16	–0.11, 0.44	0.247	0.843	0.87	–0.90, 2.68	0.337	0.651
Pastries and desserts (20 g)	–0.16	–0.54, 0.21	0.383	0.843	–1.65	–4.08, 0.84	0.194	0.523
Spreads (5 g)	–0.07	–1.19, 1.06	0.899	0.931	–1.01	–8.62, 7.23	0.802	0.928

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

High omega-3 predatory seafood intake in boys was associated with 6.96% higher urinary arsenic (95% CI: 3.50%, 10.54%). This was the only adjusted association in our analysis that remained statistically significant after FDR correction (FDR P = 0.006) (Table 7 and Supplemental Figure 5); therefore, we can say that at an FDR of q = 0.05, the association between high omega 3 predatory seafood intake and urinary arsenic in boys had an expected 0.60% of false positives. Interestingly, in girls, the intake of high omega 3 predatory seafood (–20.97%; 95% CI: –35.62%, –2.99%), eggs (–17.39%; 95% CI: –29.24%, –3.55%), and rice (–15.20%; 95% CI: –26.49%, –2.19%) were associated with lower urinary arsenic (Table 8 and Supplemental Figure 5). These associations between urinary arsenic and these food groups in girls had an expected 27.5% of false positive values at an FDR q = 0.05.

TABLE 7.

Associations between log-transformed urinary arsenic and food group intakes in boys.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.12	–0.77, 1.02	0.789	0.933	1.69	–4.69, 8.49	0.612	0.790
Homemade SSB (240 mL)	–0.16	–0.73, 0.42	0.587	0.902	–2.95	–6.97, 1.24	0.170	0.518
Homemade unsweetened drinks (240 mL)	–0.08	–0.48, 0.32	0.684	0.915	–0.22	–2.62, 2.23	0.857	0.925
Unsweetened dairy (240 mL)	–0.01	–0.74, 0.73	0.988	0.988	5.11	–0.33, 10.86	0.071	0.518
Sweetened dairy (240 mL)	0.70	–2.81, 4.35	0.695	0.915	11.06	–12.72, 41.31	0.395	0.673
High omega-3 predatory seafood (28 g)	7.47	3.85, 11.23	0.0001	0.003	6.96	3.50, 10.54	0.0002	0.006
Low omega-3 predatory seafood (28 g)	0.49	–0.13, 1.11	0.119	0.410	6.13	–1.13, 13.93	0.105	0.518
Low omega-3 nonpredatory seafood (28 g)	15.20	3.97, 27.64	0.008	0.120	65.12	–14.95, 220.57	0.143	0.518
Leafy vegetables (90 g)	2.60	0.23, 5.03	0.033	0.319	6.49	–9.15, 24.81	0.439	0.673
Root vegetables (65 g)	0.95	–0.35, 2.27	0.151	0.410	–5.56	–13.73, 3.38	0.219	0.576
Potatoes (60 g)	0.66	–3.04, 4.51	0.725	0.915	–2.36	–25.25, 27.55	0.861	0.925
Other veggies (90 g)	0.57	–0.36, 1.50	0.228	0.472	4.67	–2.00, 11.79	0.179	0.518
Fruit (90 g)	0.16	–0.13, 0.44	0.277	0.535	0.79	–1.21, 2.84	0.441	0.673
Chicken (30 g)	0.07	–0.59, 0.74	0.836	0.933	1.39	–3.05, 6.04	0.546	0.753
Eggs (50 g)	–0.07	–2.01, 1.90	0.943	0.977	–0.05	–11.77, 13.23	0.994	0.994
Soups and creams (240 mL)	–2.75	–6.44, 1.08	0.156	0.410	–0.68	–21.71, 26.00	0.955	0.989
Refined grains (30 g)	0.41	–0.23, 1.07	0.207	0.461	0.85	–3.14, 4.99	0.681	0.790
Whole grains (30 g)	0.25	–1.12, 1.64	0.720	0.915	–3.65	–12.18, 5.70	0.432	0.673
Corn (30 g)	0.29	–0.16, 0.75	0.204	0.461	2.32	–0.79, 5.54	0.150	0.518
Rice (1/4 cup or 50 g)	–0.47	–2.07, 1.16	0.564	0.902	–7.18	–16.35, 2.99	0.165	0.518
Red meat (30 g)	–0.47	–0.97, 0.04	0.070	0.407	–0.85	–4.68, 3.13	0.671	0.790
Processed meat (30 g)	0.47	0.004, 0.94	0.049	0.358	2.08	–0.77, 5.02	0.159	0.518
Legume (90 g)	0.18	–3.21, 3.69	0.916	0.977	5.19	–16.01, 31.73	0.659	0.790
Nuts (15 g)	0.37	–1.00, 1.77	0.591	0.902	6.94	–6.18, 21.90	0.317	0.673
Mexican food (150 g)	–1.19	–3.48, 1.16	0.313	0.567	9.80	–6.92, 29.54	0.270	0.653
Fast food (100 g)	–2.97	–6.24, 0.41	0.084	0.407	–10.81	–28.99, 12.01	0.327	0.673
Candy (10 g)	–0.04	–0.33, 0.26	0.805	0.933	0.85	–0.96, 2.69	0.362	0.673
Pastries and desserts (20 g)	0.24	–0.08, 0.55	0.139	0.410	0.73	–1.34, 2.84	0.492	0.713
Spreads (5 g)	1.05	–0.19, 2.32	0.098	0.407	7.26	–1.43, 16.71	0.109	0.518

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

TABLE 8.

Associations between log-transformed urinary arsenic and food group intakes in girls.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	–0.13	–1.07, 0.81	0.775	0.997	4.52	–2.22, 11.72	0.197	0.511
Homemade SSB (240 mL)	–0.06	–0.76, 0.65	0.873	0.997	0.57	–3.98, 5.33	0.811	0.940
Homemade unsweetened drinks (240 mL)	–0.02	–0.51, 0.47	0.926	0.997	–0.72	–3.66, 2.32	0.640	0.939
Unsweetened dairy (240 mL)	–0.01	–0.84, 0.83	0.986	0.997	0.91	–4.79, 6.96	0.759	0.940
Sweetened dairy (240 mL)	–1.68	–5.17, 1.94	0.354	0.997	–6.52	–27.42, 20.41	0.602	0.939
High omega-3 predatory seafood (28 g)	4.68	1.36, 8.10	0.007	0.092	–20.97	–35.62, –2.99	0.029	0.275
Low omega-3 predatory seafood (28 g)	1.81	0.27, 3.38	0.023	0.111	0.07	–9.87, 11.12	0.989	0.989
Low omega-3 nonpredatory seafood (28 g)	17.41	7.56, 28.16	0.0006	0.018	52.48	–17.29, 181.13	0.180	0.511
Leafy vegetables (90 g)	–0.86	–2.94, 1.28	0.423	0.997	2.27	–10.74, 17.19	0.746	0.940
Root vegetables (65 g)	0.54	–0.81, 1.92	0.430	0.997	–7.29	–14.83, 0.92	0.085	0.412
Potatoes (60 g)	–0.08	–3.33, 3.27	0.959	0.997	–5.56	–26.13, 20.74	0.648	0.939
Other veggies (90 g)	–0.08	–1.04, 0.89	0.871	0.997	–0.96	–6.34, 4.72	0.734	0.940
Fruit (90 g)	–0.18	–0.54, 0.18	0.318	0.997	–2.22	–4.50, 0.12	0.068	0.396
Chicken (30 g)	–0.07	–1.00, 0.86	0.879	0.997	4.55	–1.54, 11.03	0.151	0.485
Eggs (50 g)	2.85	0.56, 5.19	0.016	0.092	–17.39	–29.24, –3.55	0.019	0.275
Soups and creams (240 mL)	–2.07	–6.49, 2.56	0.369	0.997	–22.57	–43.10, 5.38	0.109	0.450
Refined grains (30 g)	0.19	–0.48, 0.86	0.578	0.997	–0.18	–4.97, 4.85	0.941	0.975
Whole grains (30 g)	–0.44	–3.18, 2.37	0.751	0.997	14.44	–8.07, 42.48	0.231	0.514
Corn (30 g)	–0.20	–0.84, 0.44	0.534	0.997	–3.06	–6.82, 0.86	0.129	0.466
Rice (1/4 cup or 50 g)	0.43	–1.44, 2.33	0.650	0.997	–15.20	–26.49, –2.19	0.028	0.276
Red meat (30 g)	–0.54	–1.53, 0.46	0.285	0.997	–6.60	–13.10, 0.38	0.068	0.396
Processed meat (30 g)	–0.11	–0.73, 0.52	0.733	0.997	1.24	–2.91, 5.57	0.563	0.939
Legume (90 g)	–4.05	–7.12, –0.88	0.014	0.092	–2.58	–24.77, 26.14	0.842	0.940
Nuts (15 g)	0.24	–2.40, 2.95	0.857	0.997	–0.87	–16.92, 18.29	0.923	0.975
Mexican food (150 g)	–3.12	–5.42, –0.76	0.011	0.092	1.91	–12.72, 19.00	0.811	0.940
Fast food (100 g)	–0.45	–4.94, 4.25	0.845	0.997	–10.67	–35.00, 22.77	0.487	0.939
Candy (10 g)	0.00	–0.28, 0.28	0.997	0.997	0.51	–1.26, 2.32	0.576	0.939
Pastries and desserts (20 g)	0.04	–0.32, 0.41	0.815	0.997	–0.93	–3.33, 1.52	0.452	0.936
Spreads (5 g)	0.39	–0.74, 1.53	0.495	0.997	–4.94	–12.14, 2.86	0.211	0.511

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

In boys, the intake of low omega-3 nonpredatory seafood was associated with 79.50% (95% CI: 3.51%, 211.29%) higher urinary cadmium, respectively. In contrast, the intake of whole grains and rice in boys was associated with lower urinary cadmium by 8.17% (95% CI: –14.86%, –0.94%) and 10.54% (95% CI: –17.84%, –2.60%), respectively (Table 9 and Supplemental Figure 6). In girls, higher homemade SSBs (240 mL), and chicken (30 g) intakes were associated with a 4.42% (95% CI: 0.13%, 8.89%), and 5.95% (95% CI: 0.38%, 11.84%) higher urinary cadmium, respectively. On the other hand, in girls, higher intakes of fruit (–2.22%; 95% CI: –4.26%, –0.12%), eggs (–16.1%; 95% CI: –27.09%, –3.38%), and nuts (–18.3%; 95% CI: –30.02%, –4.60%) were associated with lower urinary cadmium concentrations, respectively (Table 10 and Supplemental Figure 6). None of these associations between cadmium and food group intakes remained statistically significant after FDR correction, and the proportion of expected false positive values was between 38.9% and 40.6% for boys and between 26.4% and 27.8% for girls.

TABLE 9.

Associations between log-transformed urinary cadmium and food group intakes in boys.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	0.11	–0.72, 0.95	0.790	0.893	–0.40	–5.59, 5.07	0.882	0.914
Homemade SSB (240 mL)	–0.13	–0.66, 0.41	0.641	0.838	3.46	–0.07, 7.12	0.060	0.435
Homemade unsweetened drinks (240 mL)	–0.10	–0.46, 0.27	0.599	0.838	–0.49	–2.47, 1.52	0.629	0.770
Unsweetened dairy (240 mL)	0.40	–0.28, 1.09	0.248	0.562	0.14	–4.20, 4.67	0.951	0.951
Sweetened dairy (240 mL)	1.76	–1.54, 5.17	0.296	0.614	–7.30	–24.02, 13.09	0.456	0.770
High omega-3 predatory seafood (28 g)	2.35	–0.96, 5.77	0.165	0.468	7.32	–8.82, 26.31	0.397	0.770
Low omega-3 predatory seafood (28 g)	–0.16	–0.73, 0.42	0.586	0.838	1.75	–4.10, 7.96	0.566	0.770
Low omega-3 nonpredatory seafood (28 g)	9.83	–0.23, 20.91	0.057	0.288	79.50	3.51, 211.29	0.042	0.406
Leafy vegetables (90 g)	3.74	1.54, 5.99	0.001	0.037	3.72	–8.91, 18.11	0.581	0.770
Root vegetables (65 g)	1.22	–0.05, 2.50	0.060	0.288	–1.86	–8.93, 5.76	0.622	0.770
Potatoes (60 g)	–0.35	–3.76, 3.19	0.842	0.905	–5.67	–24.37, 17.65	0.605	0.770
Other veggies (90 g)	0.59	–0.27, 1.46	0.177	0.468	0.54	–0.19, 1.28	0.154	0.744
Fruit (90 g)	0.23	–0.03, 0.49	0.084	0.346	0.26	–1.39, 1.93	0.759	0.881
Chicken (30 g)	0.00	–0.61, 0.62	0.997	0.997	0.89	–2.77, 4.70	0.637	0.770
Eggs (50 g)	–0.63	–2.43, 1.19	0.486	0.830	–1.05	–10.69, 9.63	0.839	0.902
Soups and creams (240 mL)	–2.07	–5.53, 1.52	0.252	0.562	–2.19	–19.63, 19.04	0.825	0.902
Refined grains (30 g)	0.71	0.11, 1.31	0.022	0.219	–1.46	–4.68, 1.86	0.385	0.770
Whole grains (30 g)	0.16	–1.11, 1.45	0.801	0.893	–8.17	–14.86, –0.94	0.032	0.407
Corn (30 g)	0.12	–0.30, 0.55	0.575	0.838	–1.20	–3.71, 1.37	0.358	0.770
Rice (1/4 cup or 50 g)	0.06	–1.44, 1.58	0.937	0.971	–10.54	–17.84, –2.60	0.013	0.389
Red meat (30 g)	0.21	–0.26, 0.68	0.380	0.688	–1.36	–4.57, 1.95	0.416	0.770
Processed meat (30 g)	0.51	0.08, 0.95	0.023	0.219	1.45	–0.90, 3.86	0.233	0.770
Legume (90 g)	–0.63	–3.76, 2.60	0.694	0.838	–5.19	–21.24, 14.13	0.573	0.770
Nuts (15 g)	0.28	–1.04, 1.62	0.676	0.838	–2.81	–12.07, 7.42	0.577	0.770
Mexican food (150 g)	–0.56	–2.71, 1.63	0.608	0.838	4.38	–8.97, 19.69	0.540	0.770
Fast food (100 g)	2.27	–0.94, 5.59	0.166	0.468	–9.30	–24.82, 9.43	0.310	0.770
Candy (10 g)	0.12	–0.15, 0.40	0.373	0.688	1.23	–0.27, 2.75	0.114	0.664
Pastries and desserts (20 g)	0.32	0.03, 0.61	0.033	0.236	0.71	–1.00, 2.45	0.422	0.770
Spreads (5 g)	0.98	–0.18, 2.15	0.100	0.361	2.15	–4.74, 9.54	0.551	0.770

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

TABLE 10.

Associations between log-transformed urinary cadmium and food group intakes in girls.

	Crude model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	–0.45	–1.41, 0.52	0.361	0.899	0.34	–5.52, 6.56	0.912	0.944
Homemade SSB (240 mL)	–0.11	–0.84, 0.62	0.763	0.899	4.42	0.13, 8.89	0.048	0.278
Homemade unsweetened drinks (240 mL)	–0.18	–0.68, 0.33	0.487	0.899	1.04	–1.66, 3.81	0.453	0.852
Unsweetened dairy (240 mL)	–0.10	–0.96, 0.78	0.824	0.899	–1.73	–6.74, 3.56	0.515	0.852
Sweetened dairy (240 mL)	4.21	0.44, 8.13	0.030	0.435	–3.39	–22.70, 20.74	0.761	0.852
High omega-3 predatory seafood (28 g)	1.40	–2.03, 4.95	0.423	0.899	–0.18	–17.61, 20.94	0.985	0.982
Low omega-3 predatory seafood (28 g)	–1.92	–3.48, –0.34	0.019	0.435	–2.39	–11.36, 7.48	0.622	0.852
Low omega-3 nonpredatory seafood (28 g)	4.09	–5.28, 14.38	0.400	0.899	–12.58	–50.42, 54.11	0.642	0.852
Leafy vegetables (90 g)	0.23	–1.97, 2.48	0.837	0.899	–1.64	–13.09, 11.33	0.794	0.852
Root vegetables (65 g)	0.05	–1.37, 1.48	0.950	0.950	–1.11	–8.48, 6.85	0.777	0.852
Potatoes (60 g)	–0.46	–3.82, 3.03	0.791	0.899	22.97	–1.46, 53.46	0.072	0.349
Other veggies (90 g)	–0.15	–1.15, 0.86	0.770	0.899	–2.00	–6.84, 3.09	0.434	0.852
Fruit (90 g)	0.15	–0.23, 0.52	0.436	0.899	–2.22	–4.26, –0.12	0.043	0.277
Chicken (30 g)	–0.29	–1.26, 0.68	0.551	0.899	5.95	0.38, 11.84	0.041	0.277
Eggs (50 g)	0.87	–3.18, 1.52	0.471	0.899	–16.07	–27.09, –3.38	0.018	0.264
Soups and creams (240 mL)	–2.49	–7.09, 2.33	0.302	0.899	6.85	–19.26, 41.39	0.643	0.852
Refined grains (30 g)	0.18	–0.51, 0.88	0.602	0.899	–1.21	–5.39, 3.15	0.581	0.852
Whole grains (30 g)	1.16	–1.74, 4.15	0.432	0.899	–4.50	–21.88, 16.76	0.653	0.852
Corn (30 g)	0.10	–0.57, 0.76	0.772	0.899	1.05	–2.50, 4.72	0.567	0.852
Rice (1/4 cup or 50 g)	0.24	–1.69, 2.22	0.804	0.899	–2.52	–14.49, 11.13	0.702	0.852
Red meat (30 g)	–0.06	–1.10, 0.99	0.905	0.938	–4.86	–10.83, 1.50	0.136	0.394
Processed meat (30 g)	0.32	–0.33, 0.97	0.327	0.899	–1.89	–5.47, 1.83	0.318	0.838
Legume (90 g)	–0.77	–4.11, 2.69	0.654	0.899	5.44	–16.64, 33.35	0.658	0.852
Nuts (15 g)	–1.60	–4.32, 1.19	0.254	0.899	–18.29	–30.02, –4.60	0.014	0.264
Mexican food (150 g)	–0.59	–3.09, 1.97	0.642	0.899	–2.98	–15.91, 11.93	0.678	0.852
Fast food (100 g)	0.64	–4.08, 5.58	0.793	0.899	25.81	–5.27, 67.10	0.117	0.394
Candy (10 g)	0.14	–0.14, 0.43	0.329	0.899	0.26	–1.33, 1.88	0.750	0.852
Pastries and desserts (20 g)	0.24	–0.14, 0.62	0.222	0.899	1.72	–0.51, 3.99	0.135	0.394
Spreads (5 g)	0.41	–0.77, 1.60	0.495	0.899	–5.58	–11.95, 1.26	0.112	0.394

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

Several sensitivity analyses were performed to evaluate the effect of alternative food group classifications, imputation of the specific gravity values, the impact of repeated measures, residual confounding, and effect modification. The models using the alternative food group containing all fruit items regardless of presentation did not differ from the original food groupings used in our analyses (data not shown). We also fit cadmium and arsenic models adjusting for the unimputed specific gravity variable (n = 496), and no differences were observed between these models and our main models using the imputed specific gravity variable (data not shown). Similarly, values under the detection limit for urinary cadmium were replaced with the mean of these; however, the use of this variable did not alter our results (data not shown).

The main and sex-stratified arsenic models were further adjusted for the main source of drinking water at home and smoking status to account for other sources of arsenic. The drinking water source consideration did not significantly change the results (data not shown). Similarly, although urinary cadmium concentrations in adolescence differed by smoking status [yes compared with no; median (IQR): 0.08 (0.04, 0.13) μg/L compared with 0.07 (0.04, 0.11) μg/L; P = 0.041], adjusting our cadmium models by this variable did not alter the results. The girls’ models were adjusted for menarche using the available information for the peripuberty and adolescent visits, where 44.44% (n = 28) and 80.69% (n = 163) of the girls had reached menarche, respectively. The inclusion of this variable did not impact our results (data not shown). Because only 23.6% of our analytical sample had attended both visits, we performed a cross-sectional analysis only considering the observations from the adolescent visit (n = 395), with similar results to those from the main longitudinal analysis (peripuberty and adolescence visits; n = 514). Finally, to evaluate the effect modification by arsenic speciation, we stratified the arsenic models by fish and seafood intake [inorganic arsenic (<28 g/wk) compared with inorganic +organic arsenic (≥28 g/wk)] (Table 11 and Supplemental Figure 7). In nonfish or seafood consumers, the intake of leafy greens (2 cups) and chicken (100 g) was associated with 25.88% (95% CI: 8.11%, 46.58%; P = 0.011) and 5.89% (95% CI: 0.34%, 11.76%; P = 0.045) higher urinary arsenic, respectively. At an FDR q = 0.05, these associations had an expected 29.9% and 54.9% of false positive values, respectively.

TABLE 11.

Associations between log-transformed urinary arsenic and food group intake by fish/seafood intake.

	<28 g /wk				≥28 g /wk
	Adjusted model				Adjusted model
	%	95% CI	P	FDR P	%	95% CI	P	FDR P
Processed SSB (240 mL)	–1.39	–8.60, 6.40	0.678	0.904	3.08	–2.69, 9.19	0.302	0.695
Homemade SSB (240 mL)	1.41	–3.21, 6.24	0.503	0.904	–3.26	–7.21, 0.86	0.123	0.695
Homemade unsweetened drinks (240 mL)	–2.65	–5.34, 0.12	0.063	0.549	0.37	–2.32, 3.12	0.790	0.877
Unsweetened dairy (240 mL)	–0.61	–6.37, 5.51	0.817	0.904	2.93	–2.35, 8.49	0.283	0.695
Sweetened dairy (240 mL)	–3.92	–24.52, 22.30	0.708	0.904	3.89	–17.74, 31.20	0.747	0.877
High omega-3 predatory seafood (28 g)	NA	NA	NA	NA	–12.62	–26.31, 3.63	0.124	0.695
Low omega-3 predatory seafood (28 g)	NA	NA	NA	NA	5.05	–2.05, 12.67	0.170	0.695
Low omega-3 nonpredatory seafood (28 g)	NA	NA	NA	NA	61.82	–2.51, 168.58	0.067	0.695
Leafy vegetables (90 g)	25.88	8.11, 46.58	0.011	0.299	–4.97	–17.54, 9.52	0.479	0.818
Root vegetables (65 g)	–3.70	–14.49, 8.45	0.480	0.904	–6.99	–14.37, 1.03	0.090	0.695
Potatoes (60 g)	1.02	–18.79, 25.66	0.916	0.916	–13.97	–35.74, 15.16	0.311	0.695
Other veggies (90 g)	–0.71	–9.38, 8.79	0.858	0.904	1.35	–3.78, 6.75	0.611	0.877
Fruit (90 g)	–0.37	–3.41, 2.78	0.789	0.904	–0.49	–2.44, 1.49	0.622	0.877
Chicken (30 g)	5.89	0.34, 11.76	0.045	0.549	–1.81	–6.57, 3.20	0.469	0.818
Eggs (50 g)	4.61	–11.14, 23.16	0.536	0.904	–1.72	–13.58, 11.78	0.790	0.877
Soups and creams (240 mL)	3.34	–26.49, 45.28	0.826	0.904	–16.27	–34.12, 6.41	0.149	0.695
Refined grains (30 g)	0.51	–5.13, 6.48	0.842	0.904	0.03	–3.89, 4.10	0.990	0.990
Whole grains (30 g)	10.00	–6.13, 28.90	0.203	0.904	–2.46	–14.56, 11.34	0.710	0.877
Corn (30 g)	0.51	–2.94, 4.07	0.743	0.904	–0.87	–4.27, 2.65	0.620	0.877
Rice (1/4 cup or 50 g)	–0.99	–13.65, 13.53	0.869	0.904	–7.26	–16.79, 3.37	0.176	0.695
Red meat (30 g)	–1.48	–6.50, 3.81	0.524	0.904	–1.90	–6.54, 2.97	0.436	0.818
Processed meat (30 g)	1.71	–2.27, 5.85	0.353	0.904	1.76	–1.34, 4.95	0.269	0.695
Legume (1/2 or 90 g)	–4.78	–30.91, 31.24	0.729	0.904	–1.03	–20.01, 22.44	0.923	0.956
Nuts (15 g)	–3.63	–16.64, 11.4	0.567	0.904	9.58	–6.95, 29.04	0.273	0.695
Mexican food (150 g)	6.31	–11.94, 28.33	0.470	0.904	3.05	–10.92, 19.21	0.684	0.877
Fast food (100 g)	–3.69	–22.78, 20.12	0.700	0.904	–13.43	–37.07, 19.09	0.374	0.776
Candy (10 g)	–0.29	–2.02, 1.48	0.712	0.904	1.02	–0.81, 2.88	0.277	0.695
Pastries and desserts (20 g)	–0.63	–2.81, 1.59	0.521	0.904	–0.27	–2.51, 2.03	0.817	0.877
Spreads (5 g)	5.80	–3.72, 16.27	0.205	0.904	–1.23	–8.35, 6.45	0.744	0.877

Abbreviations: CI, confidence interval; FDR, false discovery rate; SSBs, sugar-sweetened beverages.

Discussion

In this study, we evaluated the associations between food group intakes and biomarker concentrations of lead (blood), arsenic (urine), and cadmium (urine) in a group of Mexican adolescents (10–18 y). Most of the observed associations were plausible according to the current literature on metal exposure sources in the food supply chain, potential metal–mineral interactions, and our understanding of cultural factors in Mexico. We found positive associations between metal exposures and food groups considered healthy or unhealthy, highlighting the need for strategies to reduce these exposures and to expand current public health recommendations. Contrarily, we observed inverse associations between metals and food groups that require further examination while considering the role of nutrients in the metabolism of metals.

The observed association between candy intake and higher blood lead concentrations is consistent with previous publications in ELEMENT, where candy intake was associated with blood lead in younger children aged 2–6 y old [57]. Similarly, studies that analyzed Mexican commercial candy found that they contained lead [21]. Although the original lead source is unclear, several factors could be at play, such as their ingredients, processing, and packaging [[58], [59], [60]]. In addition to Mexican candy, hot sauces in Mexico are also considered a leading source [61]. This is relevant as pepper and hot sauces are commonly added to fruit, raw vegetables, and other meals [62], and because commercial candies in Mexico often contain pepper-derived products [61]. Therefore, these fruit additives could partially explain the observed association between fruit intake and higher blood lead concentrations in girls. Furthermore, fruits, vegetables, and cereals can be lead sources by themselves due to plants’ uptake of lead and other metals from soil or by their direct contact with contaminated soil or water [63]. Due to their larger surface area, leafy vegetables are more vulnerable to physical metal contamination [63]. A recent review found that minimally processed food, such as root vegetables, other vegetables, and cereals, is contaminated with heavy metals around the world; notably, they observed the highest metal concentrations in leafy, brassica, and fruiting vegetables, as well as potatoes, carrots, and rice [64].

We also observed positive associations between lead, arsenic, and cadmium, and the different fish and seafood groups. These results were expected as fish and seafood are common sources of various environmental toxicants, including metals and metalloids [65]. Specifically, the low omega-3 predatory seafood group, comprising canned tuna and sardine, was positively associated with urinary arsenic. This is consistent with a recent study that analyzed 222 canned tuna samples for metal contamination which showed that total arsenic had the highest concentrations of the analyzed metals, followed by lead, mercury, and cadmium [66]. Interestingly, high omega-3 predatory seafood intake was associated with higher urinary arsenic in boys, whereas in girls it was associated with higher lead and lower arsenic concentrations. Furthermore, the intake of low omega-3 nonpredatory seafood was associated with higher blood lead and urinary cadmium concentrations in boys. The diversity of items within these groups and sex-specific dietary choices could partially explain some of the observed sex differences. The FFQ items within the high omega 3 predatory and low omega-3 nonpredatory seafood groups were nonspecific (“fresh fish” and “seafood, shrimp, or oysters,” respectively). Also, the considerable variability in pollution in the source region, species, metal, and tissue accumulation [65], and the role of selenium in arsenic excretion (discussed below) should not be disregarded.

The intake of homemade SSBs was significantly associated with cadmium concentrations, whereas the same beverages without added sugar (homemade unsweetened drinks) were not. Cadmium is a heavy metal that is released into the atmosphere, soil, and water, mainly as the result of anthropogenic activities [67]. Cadmium then enters the food supply when absorbed by diverse organisms and plants (e.g., leafy and root vegetables, cereals, and grains), including sugar cane [68]. The propensity of sugar cane for cadmium absorption and accumulation could explain the associations observed only for the homemade SSBs (which contained added sugar), compared with the homemade unsweetened drinks. Another common dietary source for cadmium is chocolate, particularly dark chocolate, as the cacao plant readily absorbs this metal and concentrates it in the cacao beans [69]. Similarly, lead has been found in cocoa products likely due to contamination of raw ingredients during manufacturing [70]. Despite this, we did not observe any associations among cadmium, lead, and our candy food group (which considered chocolate).

In our analysis stratified by fish and seafood intake, in the nonseafood and fish consumers (dietary inorganic arsenic consumers), we observed positive associations between the intake of leafy greens, chicken, and urinary arsenic. Inorganic arsenic has natural (e.g., erosion of rocks containing arsenic), as well as anthropogenic sources (e.g., mining, coal use, arsenic-based pesticides, etc.), and is absorbed by plants, such as leafy greens and grains, from contaminated water, soil, and sediments [71]. Thus, the use of arsenic and cadmium-contaminated water and food in chicken production [72] could explain the observed positive associations between chicken and arsenic, and chicken and cadmium in girls.

We also observed several inverse associations for all the metals, particularly for urinary cadmium and urinary arsenic. For example, we found an inverse association between blood lead and sweetened dairy in boys. Lead and calcium compete for the vitamin D-regulated luminal calcium transport protein 1 for absorption, and for this reason, calcium supplementation has been proposed as a feasible approach to treat lead exposure in children [73]. Our inverse associations between dairy and blood lead are consistent with other observational studies on milk, calcium, and blood lead in children [74,75]. However, randomized control trials have not been able to reproduce these results [76,77]. We found inverse associations between urinary arsenic and the intake of rice, root vegetables, high omega-3 predatory seafood, and eggs. It is unclear if these inverse associations suggest that these food items contain less arsenic in contrast to other studies [78], or if they are due to chance. Another potential explanation is the high content of selenium in seafood, eggs, and certain varieties of rice [79], and the positive correlation of root vegetables, which are low in selenium, with Mexican rice preparations. More studies are needed to confirm these findings. Similarly, some of the food groups that were inversely associated with urinary cadmium in this study (i.e., whole grain, some rice varieties, and nuts, like pumpkin seeds and peanuts) are also rich in zinc or contain zinc in smaller amounts, such as eggs [80]. Zinc and cadmium share similar chemical and physical properties and bind preferentially to the same proteins, such as albumin in the bloodstream and metallothionein (Mt). It has been suggested that increased zinc intake could reduce cadmium absorption and accumulation due to their competition for intestinal Mt [81]. The competition for absorption between zinc and cadmium has also been observed in epidemiological studies. For example, a study in NHANES 2003–2012 observed that dietary zinc intake, evaluated using 2 24-h recalls, was associated with lower urinary cadmium [82].

Among the limitations of the present analysis, we can mention the potential for residual confounding, for example, related to the variety, place of purchase, brand, and preparation of the food items we were not able to capture with the FFQ. The use of traditional lead-glazed ceramics, commonly used in food serving and preparations, is the main source of lead for Mexican children [19,20,83]. The reported use of lead-glazed ceramics during gestation in the ELEMENT cohort dropped from 16.52% in the first trimester to 11.26% in the second trimester and then to 7.36% in the third trimester [84]. Although this decrease may suggest reduced use, we cannot eliminate the possibility of social desirability bias as the children’s mothers had been advised against their use. In our analysis, we did not adjust for this variable as only 3 mothers from our analytical sample (0.75%) reported their use for their children's food storage and preparation during the first year of life, and data on their usage were not collected in subsequent study visits relevant to the current analysis. However, it is unlikely that the associations between food groups and blood lead were confounded by lead-glazed ceramics usage.

The inverse associations observed between some food groups and metal exposure biomarkers could partially be explained by the role of nutrients like calcium, selenium, and zinc in the metabolism of toxic metals. However, they are not the only ones, and confounding by other minerals or factors could exist. Similarly, our statistical analysis did not consider the correlation among food items, as we evaluated each food group and metal exposure one at a time. Metal exposures from dietary sources are interrelated as humans are not exposed to these food groups in isolation, and the intake of 1 item might affect the exposure to others. Temporality issues in these associations can arise from the discrepancy of the FFQ’s recall of 7 d compared with the longer life span of blood lead of 1–2 mo [32] and urinary cadmium of a few years to decades [34]. Finally, another important limitation of our analysis is the small sample size, and further research is needed to confirm our results. Despite this, our study has several strengths. The repeated measures of the exposure and outcomes minimize the possibility of reverse causation. Also, our study performed several sensitivity analyses that demonstrated the robustness of our results. Furthermore, FDR, a conservative P value correction, was applied to all our models to account for multiple testing.

We observed associations between metal exposures and food groups that are part of current dietary recommendations, such as fruits, vegetables, and poultry [85]. These findings evidence the challenges involved in avoiding these exposures for all age groups. On the other hand, lead content in Mexican commercial candy has been previously documented [21,57]. Although the United States FDA has a guidance of 0.1 ppm of lead level in candy, which considers Mexican candy [86], there is no formal regulation for candy producers in Mexico. There is a need for designing and developing a multisectoral strategy to control metal exposures through dietary sources, such as candy, to establish guidelines with maximum limits for toxic elements and monitoring systems ensuring compliance. We also observed a positive association between homemade SSBs and cadmium. The National Survey on Health and Nutrition from Mexico 2020–2022 reported that 62.3% of adolescents (12–19 y) exceed the current guidelines for added sugar consumption [87] (<10% of total energy intake for individuals >2 y old) [88]. Additionally, it is estimated that 9.5% of the added sugar consumed by adolescents comes from sweetened beverages [87]. Our findings on metal contamination of candy and SSBs, compounded with the high consumption of added sugars and the 42.6% prevalence of overweight or obesity in Mexican adolescents [87], highlight the need to strengthen current public health policies aimed at reducing the consumption of SSBs and other energetically dense foods in this population.

Finally, in our population, blood lead, urinary arsenic, and urinary cadmium in our population were below CDC reference values; however, the median metal concentrations in this sample were considerably higher when compared with United States adolescents (12–19 y) from NHANES 2017–2018 for blood lead (2.40 compared with 0.39 μg/dL), urinary arsenic (11.83 compared with 5.09 μg/dL) and urinary cadmium (0.09 compared with 0.06 μg/dL) [89]. The adolescent population remains an often-overlooked population despite the detrimental effects metals have in this population [1,[3], [4], [5]], and a new focus should be placed on reducing the environmental exposures in this age group, which ultimately will benefit the rest of the population.

Author contributions

The authors’ responsibilities were as follows – YR-C, AB, SKP, PXKS, CL, ER-N, KEP: designed research; KEP, MMT-R, AC, NB: conducted research; KEP, MMT-R, JDM, AC, NB: provided essential materials and databases necessary for the research; YR-C: performed statistical analysis; YR-C, AB, SKP, KEP, CL: wrote the article; and all authors: read and approved the final manuscript.

Data availability

Data described in the manuscript, code book, and analytic code will be made available on request.

Funding

This research was funded by the U.S. Environmental Protection Agency (RD83480019, RD83543601), the National Institute for Environmental Health Sciences (P20 ES018171, P01 ES02284401, P42 ES05947, R01 ES007821, R01 ES032330, and P30 ES017885), the National Institute on Aging under grant R01-AG070897 and by México Consejo Nacional de Ciencia y Tecnología under grant number 4150M9405.

Conflict of interest

The authors declare no potential conflict of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: