Abstract
Background:
Food and water are common exposure sources of arsenic and lead among children. Whereas dietary sources of these toxicants are fairly well-studied, the contribution of drinking water to toxicant exposures is not well characterized in many populations, particularly in the Global South.
Objective:
To assess the extent to which consumption of household drinking water contributes to arsenic and lead exposure among Uruguayan schoolchildren with low-level exposure.
Methods:
Children, aged 5-8 years, were enrolled into the Salud Ambiental Montevideo study during 2009-2013 from schools in Montevideo, Uruguay. Participants reported water intake as part of two 24-hour dietary recalls. Concentrations of arsenic were measured in first morning void urine samples, and adjusted for urinary specific gravity. Lead concentrations were measured in venous blood samples. Drinking water samples were collected from participants’ homes and toxicant concentrations measured. Data analyses involved a triangulation approach. First, multivariable linear regressions estimated the associations between toxicant exposure through drinking water, calculated for each child as the product of water intake and water toxicant concentration, and the respective toxicant biomarker concentrations among children with complete data on all variables (Sample A; n=40). Second, regressions were repeated for participants with complete data on all variables except water intakes (Sample B; n=195), after water intakes were imputed. Finally, models were constructed for participants of Sample B (n=195) based on drinking water intakes assumed to be fixed at 25th, 50th, 75th percentile intakes of participants in sample A.
Results:
Toxicant exposure via drinking water intake was low. The triangulation approach revealed no associations between toxicant exposure through household water intake and the respective toxicant biomarker concentrations.
Conclusion:
Studies with larger samples and repeated measures are needed to confirm these findings. Nevertheless, it appears that at low water toxicant concentrations, typical water consumption is not a major contributor to children’s exposure.
Keywords: Arsenic, Lead, Drinking water, Children, Uruguay
1. Introduction
The toxicity of arsenic (As) and lead (Pb) in humans is well documented (1, 2). Among non-occupationally exposed individuals, food and drinking water are important exposure sources (1, 2). Several international studies have characterized dietary predictors of these toxicants (3-7). It is well known that seafood is a source of As, mainly organic As compounds like arsenobetaine (5, 8), whereas rice and rice-based products are identified as sources of inorganic As (9, 10). In addition to breast milk (11, 12), several dietary predictors of Pb have been recognized, such as infant feeding formulas, baby foods, juices, and vegetables grown in urban gardens (13-15).
Evidence on the patterns in intake of water and other beverages, hydration status and their health effects has been growing in recent years (16-19). In addition to the Dietary Guidelines for Americans 2015 – 2020, a number of studies have stressed the need to replace intake of sugar sweetened beverages and other soft drinks with plain water (20-23). Several health promotion strategies in the U.S. focus on encouraging consumption of plain water among schoolchildren (24-26). Importantly, drinking water intake varies between individuals, and is influenced by factors such as the individual’s intake of other beverages including soft drinks and juices, cultural influences, and access to drinking water (27-29). Given the increasing number of policies focused toward increasing the intake of plain water and the fact that toxic effects of both As and Pb may occur in sensitive individuals even at levels below the World Health Organization (WHO) drinking water threshold of 10 μg/L when exposure occurs for prolonged periods (30-33), exposure assessment through drinking water intakes estimated at individual levels is key to understanding the role of drinking water intake as a source of toxicant exposure.
Characterizing toxicant exposure from drinking water is particularly important among children because they have relatively higher water needs than adults; children have a higher proportion of body water, and a higher body water turnover rate (27). Yet, no studies have been conducted to assess exposure to As or Pb through estimated drinking water intakes among children at an individual level. In a previous publication, we showed no association between water Pb concentrations and urinary or blood Pb concentrations among Uruguayan children, however, that study did not include individual-level estimates of drinking water intakes (34). The aim of our current study was to address this gap by examining As and Pb exposure via drinking water consumption estimated at an individual level and their association with respective biomarkers among Uruguayan schoolchildren participating in the Salud Ambiental Montevideo (SAM) study.
2. Methods
2.1. Study Design and Participant Recruitment
Salud Ambiental Montevideo is an ongoing longitudinal study conducted among children in Montevideo, Uruguay, where exposure to As, and Pb has been previously documented (35-37). Groundwater, a potential As exposure source, is not a drinking water source for most of the residents of Montevideo. Drinking water is provided to households by the state, and both As and Pb levels in drinking water are closely monitored. About half of the study sample has reported drinking tap water, and half has reported drinking a combination of bottled water and tap water (38).
The current analysis includes participants recruited in first grade of school between 2009 and 2013. A total of 358 children aged 5-8 years attending private elementary schools and their mothers were enrolled into the study; details of the recruitment process have been published previously (38). The research protocol was approved by the Institutional Review Boards at the Catholic University of Uruguay, University of the Republic of Uruguay, Pennsylvania State University, and the State University of New York at Buffalo.
2.2. Assessment of Water and Dietary Intake
Two 24-hour dietary recalls were conducted by trained nutritionists, at least two weeks apart, with the child’s mother/caregiver familiar with the child’s diet. The child contributed to the recall by filling in information about foods consumed, particularly outside of the home. The question pertaining to the amount of drinking water consumed was introduced in 2012; participants whose 24-hour dietary recalls were completed prior to 2012 were not asked about their water intake. Of the 358 participants enrolled into the study, 151 were recruited in 2012 or after, and were therefore eligible for the water intake recall. Standard measurement cups were used to aid the recall of the amount of water and other beverages consumed.
Dietary intake was also assessed as part of the 24-hour dietary recalls. Questions about the name of the meals, time and place of consumption, amounts consumed, preparation methods, recipe ingredients, brand names of commercial products, and use of vitamin and mineral supplements were asked. All foods were assigned a unique code and entered, along with the amounts consumed, into a database that contained the nutrient composition of typical Uruguayan foods and preparations. Further details of the dietary assessment methods are published elsewhere (39).
2.3. Water Collection
Water was collected by study staff in a 100 mL plastic cup from the sources indicated by the caregivers as providing drinking water (mostly kitchens taps) in the participants’ homes, according to established procedures (40). About 15 mL water was extracted from the cup using a syringe (Becton- Dickinson, Franklin Lake, NJ, USA) and passed through a 0.45 μm filter (VWR International, PA, USA) into a plastic bottle, previously rinsed with 10% HNO3 and deionized water. The water samples were immediately transferred to the Research Center, Catholic University of Uruguay and stored at 4°C until their shipment to the Pennsylvania State University within 6 months of collection.
2.4. Analysis of As and Pb in Water
Concentrations of As and Pb in drinking water samples were measured at the Materials Characterization Laboratory of the Pennsylvania State University. Prior to analysis, concentrated ultrapure nitric acid (15 – 16 N HNO3) was added to the samples to bring samples up to 1-2% nitric acid by volume. Of the 154 water samples available for analysis, As and Pb concentrations in 147 samples were analyzed using Inductively Coupled Plasma Mass Spectrometry (ICP-MS) with Collision Cell Technology (Thermo Scientific XSERIES 2, Bremen, Germany). Seven water samples were analyzed by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) (Optima 5300 DV, Perkin-Elmer). Trace elements in water (NIST 1643) were used for quality control. The limits of detection were 0.03 μg/L (As) and 0.01 μg/L (Pb) for the samples (n=147) analyzed in ICP-MS. The limits of detection of samples analyzed by ICP-AES (n=7) were 0.01 μg/L (As) and 0.1 μg/L (Pb). There were no samples below the detection limits for water As, whereas water Pb had 5.6% samples below the detection limits. The observed values were used in statistical analyses.
2.5. Urine Collection
Participants collected their first morning void urine samples in cups previously rinsed with 10% HNO3 and deionized water, and provided to families a few day prior to collection. The urine samples were transported to the Center for Research, Catholic University of Uruguay, Montevideo, on ice within the day of collection and stored at −20°C in 10 mL plastic tubes also previously rinsed with 10% HNO3 and deionized water.
2.6. Measurement of Urinary Specific Gravity
Urinary specific gravity was measured using a portable specific gravity refractometer (PAL 10S, Atago Inc, USA) on the day of urine collection. The refractometer was calibrated by pipetting 2-3 drops of tap water onto the prism surface and taking a reading. After the setting was successfully completed, 2-3 drops of urine were placed onto the prism and the displayed value was noted. Specific gravity was used to adjust As concentrations to account for the dilution of urine.
2.7. Assessment of Urinary As Concentrations
Exposure to inorganic arsenic was assessed based on the sum of urinary inorganic As and its metabolites, monomethylarsonic acid, and dimethylarsinic acid, measured using HPLC-HG-ICP-MS (HG, hydride generation, selects inorganic As and its methylated metabolites into the ICP-MS, Inductively Coupled Plasma Mass Spectrometry, Agilent 7500ce, Agilent Technologies, Tokyo, Japan) at the Karolinska Institutet, Sweden. Details of this method have been described previously (38, 41). The limit of detection was 0.1 μg/L for inorganic As (III) and monomethylarsonic acid, 0.2 μg/L for dimethylarsinic acid, and 0.3-0.5 μg/L for inorganic As (V). The intra- and inter-assay coefficients of variation were ~4%. Seven urine samples (2.1%) were below the limit of detection for inorganic As (III) and 26 (7.9%) were below the limit of detection for inorganic As (V). The measured values were used in statistical analyses.
2.8. Blood Collection
Fasting blood samples were collected by a trained phlebotomy nurse between 8:00 am and 11:00 am at the children’s schools. Approximately 3 mL venous blood was collected from each child using a 25-gauge safety butterfly blood collection set (Vacutainer, Becton Dickinson, Franklin Lakes, NJ) in heparin coated trace-metal free tubes (Vacutainer, Becton Dickinson, Franklin Lakes, NJ) for analysis of blood Pb levels.
2.9. Assessment of Blood Pb Concentrations
Blood Pb levels were measured at the Toxicology Laboratory “CEQUIMTOX” (Specialized Center for Chemical Toxicology), Faculty of Chemistry, University of the Republic of Uruguay using Atomic Absorption Spectrometry (AAS, VARIAN SpectrAA-55B) with flame or graphite furnace ionization techniques, depending on the volume of whole blood available. The graphite furnace was used in those blood samples for which volume was below 2 mL, and therefore insufficient to be measured with a flame furnace. The limit of detection was 1.8 μg/dL for the flame and 0.8 μg/dL for graphite furnace AAS techniques. Details of this procedure are published previously (41).
2.10. Parental Questionnaires, Household Crowding, and the Home Observation for Measurement of the Environment (HOME) Inventory Score
Parents completed questionnaires about the family’s sociodemographic characteristics, child’s medical history, household crowding, and the home environment. Household crowding was defined as having ≥3 persons per bedroom. A household possessions score was calculated based on parental responses to questions pertaining to 15 assets owned by the family. Responses on these items were subjected to a factor analysis. Upon oblimin rotation, the ownership of five items – computer, car, refrigerator, laundry, and a landline telephone, was used in calculating the household possessions score (3). The HOME inventory measures the developmental stimulation available in the child's home environment (42). As described previously (43), this score was based on two assessments - a home visit made by study staff, and a questionnaire filled out by the parents.
2.11. Anthropometry
Children’s height and weight were measured in triplicate by trained pediatric nurses or nutritionists using a portable stadiometer (Seca 214, Shorr Productions, Columbia, MD) and a digital scale (Seca 872, Shorr Productions, Columbia, MD), respectively, and then averaged. Body Mass Index (BMI) was calculated as weight (kg) / square of height (meters).
2.12. Statistical Analyses
Of the 358 participants enrolled into the study, 151 were eligible for the water intake recall. Of those, 40 had complete data on drinking water concentrations of both toxicants, toxicant biomarkers, sociodemographic, anthropometric, and biochemical covariates of interest, and had at least one reported water intake volume (18 participants had water intake values reported in both recalls, 14 had reported in recall 1, 8 had reported in recall 2). This complete case sample is henceforth referred to as sample A (n=40). Water intake values obtained from the two 24-hour dietary recalls were averaged for participants who had reported intakes in both recalls (r=0.6 for intakes reported in both recalls). For participants with only one completed recall, that value was used in analyses. Additionally, 155 of 358 participants had complete data on all the variables of interest, except for water intake levels. These 155, in addition to the complete case sample, is henceforth referred to as sample B (n=155+40=195).
Statistical analyses were performed using SAS version 9.4. As a first step, descriptive analyses were conducted to assess the sociodemographic, anthropometric, and biochemical characteristics of participants in samples A and B. Next, toxicant exposure through drinking water (μg/day) was calculated as the product of the individual participant’s daily water intake (L) and water concentration of each toxicant (μg/L) among participants of sample A. Separate multivariable linear regression models were constructed in sample A with As and Pb exposure via water intake as predictors and the respective toxicant biomarker (urinary As and blood Pb) as the outcomes; models were adjusted for age, sex, BMI, maternal education, intake of milk and soft drinks, and year of enrollment into the study. Models containing water As and urinary As were further adjusted for the consumption of rice, a known source of As exposure, to facilitate the estimation of the independent association between As exposure via drinking water intake and urinaryAs concentrations. Covariates for inclusion in the regression models are known predictors of As and Pb exposure markers and were identified a priori.
Next, multiple imputations with chained equations were carried out using STATA version 12 to impute missing water intake values for 155 participants in sample B, based on the reported water intake levels in sample A. A series of 50 iterations were completed; water intake values were imputed separately for the two 24-hour dietary recalls and then the imputed intake values from both recalls were averaged. Variables included in the imputation model were sex (male/female), age (continuous, months), maternal education (continuous, number of years), BMI (continuous, kg/m2), household possessions score (continuous), household crowding (yes/no, defined as ≥3 persons per bedroom), HOME inventory score (continuous), hemoglobin levels (continuous, g/dL), intakes of milk and soft drinks (continuous, mL/day), rice intake (continuous, grams/day), and total energy intake (continuous, kcal/day). Then, multivariable linear regression analyses were performed, similar to those in sample A. As part of model diagnostics, the values of the Fraction of Missing Information (FMI), which indicate the amount of sampling variance that may be attributable to missing data, were evaluated.
To further understand the potential ranges of As and Pb exposure through drinking water and to corroborate the results observed in analyses conducted in samples A and B, drinking water intake among participants in sample B was assumed to be fixed at the 25th, 50th, and 75th percentile of intake distributions observed in sample A. The intake of As and Pb through drinking water (μg/day) was calculated as the product of these assumed intakes (L) and the measured As and Pb concentrations (μg/L) in each child’s household drinking water. Multivariable regression analyses mirrored those described above for sample A.
Additional analyses were conducted to compare (i) participant characteristics in Sample B with missing water intake data (n=155) and those with reported water intakes (n=40), (ii) water toxicant and toxicant biomarker concentrations among study participants by year of enrollment, (iii) participant characteristics prior to making exclusions based on missing covariates, between all enrolled participants with missing water intake data (n=286) to those with reported water intakes (n=72), (iv) characteristics of participants who were excluded from analysis (n=163) because of missing covariate data and those who were included in the analysis (n=195), and (v) findings of the analyses conducted in Samples A and B to those in an increased sample of 225 participants after imputing data on urinary arsenic concentrations, water arsenic concentrations, and HOME score, since these variables had large amounts of missing information. Finally, drinking water intakes among Uruguayan schoolchildren were compared to those of 5-8-year-old participants of the 2009-2013 cycles of the National Health and Nutrition Examination Survey (NHANES) to understand the drinking water intake ranges of early school-age children.
3. Results
The flowchart of the process used to select participants into the analyses is presented in Supplemental Figure 1. The sociodemographic, anthropometric, and biochemical characteristics of participants in Sample A and B are shown in Table 1. For Sample A, the median (5th percentile, 95th percentile) age was 79.5 (72.0, 92.0) months, and ~ 38% of the participants were boys; the age and sex distribution in sample B was similar to that of sample A. The median (5th percentile, 95th percentile) concentrations of urinary As and blood Pb in sample A were 14.2 (3.42, 36.0) μg/L and 3.00 (2.60, 7.35) μg/dL, respectively. Similar biomarker concentrations were observed among participants belonging to sample B. The median (5th percentile, 95th percentile) concentrations of As and Pb in the drinking water of sample A participants were 0.57 μg/L (0.23, 3.84) and 0.71 μg/L (0.11, 6.08), respectively, similar to those in sample B.
Table 1:
Sociodemographic, anthropometric, and biochemical characteristics of participants in sample Aa (n=40) and sample B (n=195) of the study.
| Variable | Sample A (n=40) | Sample B (n=195) |
|---|---|---|
| Age, months | ||
| Median (5th, 95th percentile) | 79.5 (72.0, 92.0) | 81.0 (72.0, 92.0) |
| Sex | ||
| Boys, n (%) | 15 (37.5) | 75 (38.5) |
| Girls, n (%) | 25 (62.5) | 120 (61.5) |
| BMI, kg/m2 | ||
| Median (5th, 95th percentile) | 16.5 (14.2, 22.3) | 16.4 (14.0, 22.1) |
| Maternal Education, years | ||
| 4, n (%) | 0 (0.00) | 2 (1.03) |
| 6, n (%) | 7 (17.5) | 29 (14.9) |
| 8, n (%) | 14 (35.0) | 78 (40.0) |
| 9, n (%) | 5 (12.5) | 14 (7.18) |
| 11, n (%) | 7 (17.5) | 31 (15.9) |
| 12, n (%) | 5 (12.5) | 18 (9.23) |
| 14, n (%) | 2 (5.00) | 15 (7.69) |
| 17, n (%) | 0 (0.00) | 8 (4.10) |
| Household Crowding | ||
| <3 persons/bedroom, n (%) | 35 (87.5) | 155 (79.5) |
| ≥3 persons/bedroom, n (%) | 5 (12.5) | 40 (20.5) |
| Household Possessions Score | ||
| ≥Median of 4, n (%) | 15 (37.5) | 74 (38.0) |
| <Median of 4, n (%) | 25 (62.5) | 121 (62.1) |
| HOME Inventory Score | ||
| Median (5th, 95th percentile) | 45.5 (40.5, 52.2) | 45.5 (40.4, 50.6) |
| Year of Enrollment | ||
| 2009, n (%) | 0 (0) | 38 (19.5) |
| 2010, n (%) | 0 (0) | 60 (30.8) |
| 2011, n (%) | 0 (0) | 13 (6.7) |
| 2012, n (%) | 16 (40.0) | 36 (18.5) |
| 2013, n (%) | 24 (60.0) | 48 (24.6) |
| Water intake, mL/day b | ||
| Median (5th, 95th percentile) | 262 (100, 950) | 337 (100, 925) |
| Milk intake, mL/day | ||
| Median (5th, 95th percentile) | 275 (0.00, 519) | 300 (0.00, 700) |
| Soft drink intake, mL/day | ||
| Median (5th, 95th percentile) | 0.00 (0.00, 50.0) | 0.00 (0.00, 100) |
| Rice intake, grams/day | ||
| Median (5th, 95th percentile) | 0.00 (0.00, 60.0) | 0.00 (0.00, 60.0) |
| Hemoglobin, μg/dL | ||
| Median (5th, 95th percentile) | 13.1 (11.1, 15.3) | 13.1 (11.5, 15.3) |
| Urinary Arsenic, μg/L | ||
| Median (5th, 95th percentile) | 14.2 (3.42, 36.0) | 12.0 (5.57, 32.7) |
| Blood Lead Level, μg/dL | ||
| Median (5th, 95th percentile) | 3.00 (2.60, 7.35) | 3.80 (0.80, 8.20) |
| Water arsenic, μg/L | ||
| Median (5th, 95th percentile) | 0.57 (0.23, 3.84) | 0.42 (0.16, 1.00) |
| Water lead, μg/L | ||
| Median (5th, 95th percentile) | 0.71 (0.11, 6.08) | 0.86 (0.14, 14.6) |
| Arsenic exposure via drinking water intake, μg/day b | ||
| sMedian (5th, 95th percentile) | 0.15 (0.03, 2.45) | 0.14 (0.03, 0.60) |
| Lead exposure via drinking water intake, μg/day b | ||
| Median (5th, 95th percentile) | 0.21 (0.02, 4.35) | 0.32 (0.03, 6.18) |
Abbreviations: BMI: Body Mass Index; HOME: Home Observation for Measurement of the Environment
For participants with only one 24-hour dietary recall, the water intake from that recall is reported; for those with two 24-hour dietary recalls, the average water intake volume from both recalls is reported.
Water intake values in sample B are imputed from the measured intakes in sample A
Figures 1a and 1b present the toxicant concentrations in drinking water samples collected from homes of participants in samples A and B.
Figure 1a:
Drinking water arsenic concentrations for Sample A and Sample B. As a reference, the World Health Organization recommended limit for drinking water arsenic is 10 μg/L; the same is the Uruguayan regulatory limit.
Figure 1b:
Drinking water lead concentrations for Sample A and Sample B. As a reference, the World Health Organization recommended limit for drinking water lead is 10 μg/L; the same is the Uruguayan regulatory limit.
Figure 2 presents the exposure to As and Pb respectively, calculated from measured drinking water intakes in sample A.
Figure 2:
Arsenic and lead exposure from drinking water intake in Sample A (n=40), where exposure is calculated as the product of the drinking water toxicant concentration and drinking water intake volume.
Table 2 presents the results of the multivariable linear regression analyses testing the associations between toxicant exposure through drinking water intakes and respective toxicant biomarker concentrations among participants of sample A. No evidence of an association was observed, either in the crude or covariate-adjusted models.
Table 2:
Associations between toxicant exposure through drinking water intakes (μg/day)a and respective toxicant biomarker concentrations (μg/L or μg/dL) among participants of sample A (n=40).
| Crude β (95% CI) |
Adjustedb β (95% CI) |
|
|---|---|---|
| Urinary Arsenic, μg/L | 0.33 (−2.69, 3.36) | 0.85 (−2.64, 4.34) |
| Blood Lead, μg/dL | 0.12 (−0.18, 0.43) | 0.07 (−0.29, 0.43) |
Abbreviation: CI – confidence interval
Exposure calculated as the product of the drinking water toxicant concentration and drinking water intake volume
Adjusted for age, sex, body mass index, maternal education, intake of milk, soft drinks, and year of enrollment into the study; arsenic model additionally adjusted for rice intake.
Among participants of sample B, drinking water intake values were imputed based on reported intakes of sample A. Table 3 presents the results of the multivariable linear regression analyses testing the associations between toxicant exposure through drinking water intakes and respective toxicant biomarker concentrations among these participants. As in the complete case sample, there was no evidence for an association between As or Pb exposure through drinking water and the respective biomarker concentrations. The FMI values, representing the amount of sampling variance attributable to missing data, were ranged 0.11 – 0.34, indicating efficient imputations. The largest was found for the crude model assessing the association between exposure to Pb through drinking water and blood Pb concentrations.
Table 3:
Associations between toxicant exposure through drinking water intakes (μg/L)a and respective toxicant biomarker concentrations (μg/L or μg/dL) among participants of sample B (n=195).
| Crude β (95% CI) |
Adjustedb β (95% CI) |
|
|---|---|---|
| Urinary Arsenic, μg/L | 1.17 (−1.45, 3.78) | 0.85 (−1.41, 3.10) |
| Blood Lead, μg/dL | 0.12 (−0.01, 0.25) | 0.09 (−0.03, 0.22) |
Abbreviation: CI – confidence interval
Exposure calculated as the product of the drinking water toxicant concentration and drinking water intake volume
Adjusted for age, sex, body mass index, maternal education, intake of milk, soft drinks, and year of enrollment into the study; arsenic model additionally adjusted for rice intake.
Table 4 presents the results of the multivariable linear regression analyses testing the associations between toxicant exposure through drinking water when the intake was fixed at 25th, 50th, and 75th percentile, and respective toxicant biomarker concentrations among participants in sample B. The 25th, 50th, and 75th percentile intake levels were 167 mL/day, 288 mL/day, and 550 mL/day respectively, and were obtained from measured intakes of sample A. There was no evidence of an association between As exposure through drinking water and urinary As concentrations. Lead exposure through drinking water showed a statistically significant, positive, but modest association with blood Pb concentrations at all percentile levels in crude models; the associations were no longer statistically significant upon adjusting for covariates.
Table 4:
Associations between toxicant exposure through drinking water intake fixed at 25th, 50th, and 75th percentiles (μg/L)a, and respective toxicant biomarker concentrations (μg/L or μg/dL) among participants of sample B (n=195).
| 25th percentile intake (167 mL/day) |
50th percentile intake (288 mL/day) |
75th percentile intake (550 mL/day) |
||||
|---|---|---|---|---|---|---|
| Crude β (95% CI) |
Adjustedb β (95% CI) |
Crude β (95% CI) |
Adjustedb β (95% CI) |
Crude β (95% CI) |
Adjustedb β (95% CI) |
|
| Arsenic | 2.52 (−2.76, 7.79) |
1.68 (−3.66, 7.02) |
1.46 (−1.60, 4.52) |
0.97 (−2.12, 4.07) |
0.76 (−0.84, 2.37) |
0.51 (−1.11, 2.13) |
| Lead |
0.36
(0.10, 0.63) |
0.26 (−0.01, 0.53) |
0.21
(0.06, 0.36) |
0.15 (−0.00, 0.31) |
0.11
(0.03, 0.19) |
0.08 (−0.00, 0.16) |
Abbreviation: CI – confidence interval
Percentiles obtained from intakes observed in sample A
Exposure calculated as the product of the drinking water toxicant concentration and drinking water intake volume
Adjusted for age, sex, body mass index, maternal education, intake of milk, soft drinks, and year of enrollment into the study; arsenic model additionally adjusted for rice intake.
Bolded values indicate statistical significance at an alpha level of 0.05
Supplemental Table 1 presents the tap water intake levels among Uruguayan study participants and children of comparable ages from NHANES cycles 2009-2013. Drinking water intakes in both samples were very similar, with medians <350 mL/day, indicating low intake levels. Supplemental Tables 2-5 show the comparison of covariates between various subgroups of study participants. Supplemental Table 6 presents findings of analyses conducted in an increased sample of 225 participants after imputing data on urinary arsenic concentrations, water arsenic concentrations, and HOME score; results remained unchanged.
4. Discussion
Among 5–8-year-old Uruguayan children exposed to low levels of As and Pb, we estimated toxicant exposure through drinking water based on water intake volumes reported via 24-hour dietary recalls and toxicant concentrations measured in participants’ household drinking water sources. We implemented a triangulation approach, based on study samples with complete data, imputed data on water intake, and water intake levels assumed to be fixed at the 25th, 50th, and the 75th intake percentiles of the complete case sample. In all three approaches, we found no evidence of a relationship between estimated As and Pb exposure via water intake volumes and the respective toxicant biomarker concentrations.
Several previous studies have evaluated the relationships of average concentrations of toxicants in drinking water sources with measured toxicant biomarker concentrations among study participants. For example, a study in Thailand showed a positive association between groundwater As concentrations and urinary As concentrations among adults; agrochemical usage in the study area was high, which could have contributed to toxicants entering the groundwaters (44). Among Bangladeshi children, average water As concentrations were seen to gradually decrease from 2003 to 2013, but urinary As concentrations continued to remain elevated, indicating that As exposure was likely occurring from dietary sources (45). A study among American children revealed that those drinking water from private wells had higher blood Pb levels compared to those relying on the community water systems (46). Although such studies provide consistent evidence of the role of drinking water as a toxicant exposure source, they did not include individual-level water intakes of participants.
To our knowledge, ours is one of the first studies to assess the associations between As and Pb exposure through individual-level drinking water intakes and the respective biomarker concentrations among children. Existing studies that evaluate the role of drinking water as a predictor of toxicant exposure largely focus on adults or do not include individual-level water intakes. For example, the Health Effects of Arsenic Longitudinal Study of Bangladesh consisted of a cohort of approximately 12,000 adults of whom 55% had been consuming water with As concentrations > 55 μg/L (legal limit for Bangladesh), and 76.2% of whose drinking water contained As levels >10 μg/L (WHO recommended limit) (47). In that study, a cumulative As exposure index was calculated as the product of the As concentrations in the drinking water source, the water intake volume/day, and the duration of water intake from that water source. The daily drinking water intake and the duration of well water use were not associated with urinary As concentrations; the cumulative As exposure index was not correlated with urinary As concentrations (range of coefficients: 0.30 – 0.66; all p values>0.10) (47). A study among pregnant women in New Hampshire evaluated the intake of household drinking water and rice, both measured using 3-day dietary records, as predictors of As exposure (48). Arsenic exposure through drinking water was calculated as the product of drinking water intake and drinking water As concentrations; As exposure through drinking water was positively associated with urinary As concentrations [β (95% CI): 0.03 (0.02, 0.04)]. The median (range) As exposure via drinking water was 0.27 μg/day (0.00, 133) (48). In comparison, our study participants had low levels of As exposure through drinking water intakes, median (range): 0.15 μg/day (0.01, 5.23). However, direct comparisons with findings from either of the aforementioned studies cannot be made because our study primarily focused on children, whose water consumption patterns may differ from those of adults.
Some studies have documented potential ranges of toxicant exposure by assessing water toxicant concentrations and self-reported drinking water intakes, also among adults. A cross-sectional study among Malaysian adults included the collection of first-flushed (first water sample collected in the morning, after overnight standing in pipes) and fully-flushed (water allowed to run for 2 minutes in the morning prior to sample collection) samples of water from kitchen taps to evaluate drinking water Pb concentrations and potential exposure through self-reported drinking water (49). The mean (SD) chronic daily intake of Pb was estimated at 0.03 (0.03) μg/day/kg (49). The National Human Exposure Assessment Survey evaluated temporal trends in drinking water intakes as well as drinking water concentrations of As and Pb measured six times over one year (50). The annual mean (SD) intake of As and Pb through drinking water was 0.55 μg/day (0.49) and 0.95 μg/day (1.36), respectively (50). Our study of children shows comparable levels of toxicant exposure via household drinking water; the mean (SD) exposure to As and Pb among Sample A participants was 0.43 (0.99) ug/day and 0.80 (1.73) ug/day, respectively.
In evaluating the role of drinking water in toxicant exposure among children, it is important to consider the variations in the intake of water and other beverages. The median (range) of reported drinking water intake in our study was 250 mL/day (20, 1500) and 288 mL/day (50, 1300) in the first (n=32) and second (n=26) 24-hour dietary recall, respectively. Comparable drinking water intakes were reported for 5–8-year-old participants of the NHANES cycles 2009 - 2013, the time frame in which our study was conducted (Supplemental Table 1). According to the Institute of Medicine, the Adequate Intake levels for water among 4-8-year-olds is 1.7 L/day of total water, with approximately 1.2 L as total beverages, including drinking water, and the rest as water obtained through foods (51). Our study accounted for intake of milk and soft drinks in the models that imputed drinking water intakes among 155 participants of Sample B. The majority of participants (98%) reported zero intake of juices or smoothies. Also, very few participants reported intake of water-rich fruits: 98.9% reported no intake of watermelon, cantaloupe (99.5%), kiwis (99.5%), or oranges (86.0%). We were also unable to take into account the amount of water used in cooking. Taking into account the intake of water through foods and other beverages will help in further understanding the role of water as a whole in As and Pb exposure; further studies are needed on this topic.
In our study, the prompt pertaining to water intake was introduced into the 24-hour dietary recalls in 2012. Hence, water intake data were missing among participants who were enrolled prior to 2012, or those who were enrolled 2012 onward but did not respond to that query. We compared participant characteritstics between those with missing data on water intake and those who had reported their water intakes. Results from Supplemental Table 2 indicate that although some covariates differed between the groups, there was no systematic pattern observed in missingness. Further, to evaluate potential temporal trends in water toxicant concentrations as well as toxicant biomarker concentrations, we compared these variables by year of study enrollment. We observed no distinct temporal pattern in water toxicant concentrations or toxicant biomarker levels (Supplemental Table 3). We also observed no distinct pattern of participant characteristics when we compared them based on their missing vs. non-missing water intake data in the whole study sample prior to making any exclusions for complete case analyses (n=358; Supplemental Table 4), and when we compared sociodemographic, anthropometric, and biochemical characteristics between participants who were excluded from analysis (n=163) because of missing covariate data and those who were included in the analysis (n=195); supplemental table 5. Both samples A and B were drawn from the same population. Further, with no notable differences between those who were asked about their water intake as part of the 24-hour dietary recalls but did not report them, those who actually reported intakes, and those who were not asked about their water intakes, we determined that imputation procedures could be applied to the entirety of participants within Sample B.
Our study involved a triangulation approach to analyses, wherein we assessed the associations between toxicant exposure through drinking water intakes and the respective toxicant biomarker concentrations in Sample A (n=40), imputing water intake volumes for 155 participants in Sample B (n=195), and assuming drinking water intakes fixed at specific percentiles in Sample B (n=195). Using multiple imputations with chained equations, we imputed ~80% data on the water intake variables in Sample B. Because of the high proportion of imputations, our model diagnostics consisted of the FMI statistic, which evaluates the fraction of the total variance that is attributable to between-imputation variance (52). FMI values range from 0 to 1, where a lower FMI value indicates lower levels of variability between imputed datasets and that the covariates in the imputation model are contributing information towards the imputation of missing values, thereby increasing model efficiency (52). For our multiple imputation models, all FMI values were less than 0.34, indicating efficient imputations. We ran 50 iterations which were well above the number of iterations, 34, recommended based on our highest FMI value. Despite imputing large proportions of exposure data, the validity of our study findings is maintained because of our imputation model diagnostics and the fact that we observed consistent results in the analyses conducted in Sample A (n=40), as well as in those assuming fixed drinking water intakes in Sample B (n=195).
Our study findings need to be interpreted in light of some limitations. First, our study focused on household drinking water only, while participants could have consumed water from other sources, such as bottled water, school taps, etc. Thus, it is possible that our results are an underestimation or an overestimation of the true impact of drinking water intake on toxicant biomarkers, depending on the intake levels and toxicant concentrations in these other drinking water sources. Second, our complete case sample had a modest sample size. However, we conducted multiple imputations with appropriate model diagnostics and used a triangulation approach in our analyses to address this limitation. Third, water intake in our study was assessed using 24-hour dietary recalls and toxicant concentrations in drinking water sources were based on a single sample. Twenty four-hour dietary recalls may not represent usual water intake levels. We used data collected from two 24-hour dietary recalls which were conducted at least fifteen days apart, which alleviates this limitation to some extent. Using single samples to determine certain toxicant concentrations, Pb concentrations for example, is problematic because water Pb concentrations consist of particulate Pb and soluable Pb in flowing systems. A single sample may not capture particulate Pb on that particular day, which could result in the water Pb levels being representative only of soluable Pb on that day. Fourth, we did not take into account the quantity of water used in cooking because these data were not collected as part of the 24-hour dietary recalls. Again, this indicates the potential that our results are an underestimation or an overestimation of the true exposure-outcome relationship. Fifth, we used blood Pb as the biomarker for Pb exposure, which reflects an exposure period of the preceding 2-3 months (53). Thus, this biomarker reflects an exposure window that potentially captures exposure sources other than the water intakes measured via the 24-hour dietary recalls. Sixth, exposure to both As and Pb is influenced by several factors such as inhalation and ingestion of Pb in dust and soil, proximity to emission sources, air quality etc. We were unable to account for these in our study because we did not have data on them. Finally, the cross-sectional nature of our study does not support drawing any causal conclusions.
Our study has several strengths. Urine and blood samples in our study were collected on the day of the first dietary recall, which is well suited to assess associations with short term markers of exposure, such as urinary As which is the gold standard for As exposure assessment and is particularly useful in reflecting short-term exposure (54). Further, our multiple imputation models included intake of other beverages commonly consumed by children, such as milk and soda; intake of these beverages impacts the amount of water consumed. We observed similar water intake levels in comparison to those of similarly-aged NHANES participants enrolled in the same time frame, which increases the confidence in our findings despite the modest sample size. Finally, our analyses consisted of a triangulated approach and comparing findings obtained from the three methods.
5. Conclusion
Among 5–8-year-old Uruguayan children, the exposure to As and Pb via drinking water intake was not associated with biomarker levels of these toxicants in blood or urine, either due to low levels of these toxicants in the water or low level of daily water consumption. Further, longitudinal studies with repeated measures are needed to confirm our findings.
Supplementary Material
Acknowledgements:
We thank the field personnel for help with data collection: Delma Ribeiro and Graciela Yuane collected and processed biological samples; Valentina Baccino, Elizabeth Barcia, Soledad Mangieri, Virginia Ocampo collected dietary recalls; Martín Bidegaín assisted with family and school contacts. We also thank all the study participants and their families for their valuable time.
Funding sources:
This work was supported by the National Institutes of Health, the Fogarty International Center (ES019949, PI: Kordas and ES016523, PI: Kordas), and the University at Buffalo’s Community of Excellence in Global Health Equity.
Footnotes
Conflicts of Interest: The authors declare no conflicts of interests
References
- 1.IARC. ARSENIC, METALS, FIBRES, AND DUSTS. ARSENIC AND ARSENIC COMPOUNDS. France: 2012. [PMC free article] [PubMed] [Google Scholar]
- 2.Anttila A, Apostoli P, Bond JA, Gerhardsson L, Gulson BL, Hartwig A, et al. IARC monographs on the evaluation of carcinogenic risks to humans: Inorganic and organic lead compounds. 2006. [PMC free article] [PubMed] [Google Scholar]
- 3.Kordas K, Burganowski R, Roy A, Peregalli F, Baccino V, Barcia E, et al. Nutritional status and diet as predictors of children's lead concentrations in blood and urine. Environment International. 2018;111:43–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kordas K, Queirolo EI, Mañay N, Peregalli F, Hsiao PY, Lu Y, et al. Low-level arsenic exposure: Nutritional and dietary predictors in first-grade Uruguayan children. Environmental research. 2016;147:16–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Navas-Acien A, Francesconi KA, Silbergeld EK, Guallar E. Seafood intake and urine concentrations of total arsenic, dimethylarsinate and arsenobetaine in the US population. Environmental research. 2011;111(1):110–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Stanek K, Manton W, Angle C, Eskridge K, Kuehneman A, Hanson C. Lead consumption of 18-to 36-month-old children as determined from duplicate diet collections: nutrient intakes, blood lead levels, and effects on growth. Journal of the American Dietetic Association. 1998;98(2):155–8. [DOI] [PubMed] [Google Scholar]
- 7.Xue J, Zartarian V, Wang S-W, Liu SV, Georgopoulos P. Probabilistic modeling of dietary arsenic exposure and dose and evaluation with 2003–2004 NHANES data. Environmental Health Perspectives. 2010;118(3):345–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Taylor V, Goodale B, Raab A, Schwerdtle T, Reimer K, Conklin S, et al. Human exposure to organic arsenic species from seafood. Science of the Total Environment. 2017;580:266–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Signes-Pastor AJ, Carey M, Meharg AA. Inorganic arsenic in rice-based products for infants and young children. Food chemistry. 2016;191:128–34. [DOI] [PubMed] [Google Scholar]
- 10.Carbonell-Barrachina AA, Wu X, Ramirez-Gandolfo A, Norton GJ, Burlo F, Deacon C, et al. Inorganic arsenic contents in rice-based infant foods from Spain, UK, China and USA. Environmental pollution (Barking, Essex : 1987). 2012;163:77–83. [DOI] [PubMed] [Google Scholar]
- 11.Anderson HA, Wolff MS. Environmental contaminants in human milk. Journal of exposure science & environmental epidemiology. 2000;10(6):755–60. [DOI] [PubMed] [Google Scholar]
- 12.Ettinger AS, Roy A, Amarasiriwardena CJ, Smith D, Lupoli N, Mercado-García A, et al. Maternal blood, plasma, and breast milk lead: lactational transfer and contribution to infant exposure. Environmental health perspectives. 2014;122(1):87–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gardener H, Bowen J, Callan SP. Lead and cadmium contamination in a large sample of United States infant formulas and baby foods. Science of the Total Environment. 2019;651:822–7. [DOI] [PubMed] [Google Scholar]
- 14.McBride MB, Shayler HA, Spliethoff HM, Mitchell RG, Marquez-Bravo LG, Ferenz GS, et al. Concentrations of lead, cadmium and barium in urban garden-grown vegetables: the impact of soil variables. Environmental pollution (Barking, Essex : 1987). 2014;194:254–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Reports C. Consumer Reports Finds Concerning Levels of Heavy Metals in Popular Fruit Juices 2019 [Available from: https://www.consumerreports.org/media-room/press-releases/2019/01/consumer_reports_finds_concerning_levels_of_heavy_metals_in_popular_fruit_juices/.
- 16.Maughan R. Impact of mild dehydration on wellness and on exercise performance. European journal of clinical nutrition. 2003;57(2):S19–S23. [DOI] [PubMed] [Google Scholar]
- 17.Riebl SK, Davy BM. The hydration equation: Update on water balance and cognitive performance. ACSM's health & fitness journal. 2013;17(6):21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Nissensohn M, Sánchez-Villegas A, Galan P, Turrini A, Arnault N, Mistura L, et al. Beverage consumption habits among the European population: Association with total water and energy intakes. Nutrients. 2017;9(4):383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vieux F, Maillot M, Rehm CD, Barrios P, Drewnowski A. Trends in tap and bottled water consumption among children and adults in the United States: analyses of NHANES 2011–16 data. Nutrition journal. 2020;19(1):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Committee DGA. Dietary guidelines for Americans 2015-2020: Government Printing Office; 2015. [Google Scholar]
- 21.Piernas C, Barquera S, Popkin BM. Current patterns of water and beverage consumption among Mexican children and adolescents aged 1–18 years: analysis of the Mexican National Health and Nutrition Survey 2012. Public health nutrition. 2014;17(10):2166–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Iglesia I, Guelinckx I, De Miguel-Etayo PM, González-Gil EM, Salas-Salvadó J, Kavouras SA, et al. Total fluid intake of children and adolescents: cross-sectional surveys in 13 countries worldwide. European journal of nutrition. 2015;54(2):57–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Senterre C, Dramaix M, Thiébaut I. Fluid intake survey among schoolchildren in Belgium. BMC Public Health. 2014;14(1):1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Service UFaN. Healthy hunger-free kids act. 2013.
- 25.Service. UFaN. National School Lunch Program.
- 26.Service UFaN. School Breakfast Program. [Google Scholar]
- 27.Sichert-Hellert W, Kersting M, Manz F. Fifteen year trends in water intake in German children and adolescents: results of the DONALD Study. ACTA Paediatrica. 2001;90(7):732–7. [PubMed] [Google Scholar]
- 28.EFSA Panel on Dietetic Products N, Allergies. Scientific opinion on dietary reference values for water. EFSA Journal 2010;8(3):1459. [Google Scholar]
- 29.Vieux F, Maillot M, Constant F, Drewnowski A. Water and beverage consumption patterns among 4 to 13-year-old children in the United Kingdom. BMC Public Health. 2017;17(1):1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Organization WH. Lead in drinking-water: background document for development of WHO guidelines for drinking-water quality. World Health Organization; 2003. [Google Scholar]
- 31.Gadd GM. Arsenic Pollution: A Global Synthesis. By P. Ravenscroft, H. Brammer and K. Richards. Chichester, UK: Wiley-Blackwell,(2009), pp. 588.£ 65.00 (paperback). ISBN 978-1-4051-8602-5. Experimental Agriculture. 2009;45(4):509-. [Google Scholar]
- 32.Weitzman M. Blood lead screening and the ongoing challenge of preventing children’s exposure to lead. JAMA Pediatrics. 2019;173(6):517–9. [DOI] [PubMed] [Google Scholar]
- 33.Bräuner EV, Nordsborg RB, Andersen ZJ, Tjønneland A, Loft S, Raaschou-Nielsen O. Long-term exposure to low-level arsenic in drinking water and diabetes incidence: a prospective study of the diet, cancer and health cohort. Environmental Health Perspectives. 2014;122(10):1059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ravenscroft J, Roy A, Queirolo EI, Mañay N, Martínez G, Peregalli F, et al. Drinking water lead, iron and zinc concentrations as predictors of blood lead levels and urinary lead excretion in school children from Montevideo, Uruguay. Chemosphere. 2018;212:694–704. [DOI] [PubMed] [Google Scholar]
- 35.Kordas K, Queirolo EI, Ettinger AS, Wright RO, Stoltzfus RJ. Prevalence and predictors of exposure to multiple metals in preschool children from Montevideo, Uruguay. Science of the Total Environment. 2010;408(20):4488–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Mañay N, Cousillas AZ, Alvarez C, Heller T. Lead contamination in Uruguay: the “La Teja” neighborhood case. Reviews of Environmental Contamination and Toxicology: Springer; 2008. p. 93–115. [PubMed] [Google Scholar]
- 37.Queirolo EI, Ettinger AS, Stoltzfus RJ, Kordas K. Association of anemia, child and family characteristics with elevated blood lead concentrations in preschool children from Montevideo, Uruguay. Archives of Environmental & Occupational Health. 2010;65(2):94–100. [DOI] [PubMed] [Google Scholar]
- 38.Desai G, Barg G, Queirolo EI, Vahter M, Peregalli F, Mañay N, et al. A cross-sectional study of general cognitive abilities among Uruguayan school children with low-level arsenic exposure, potential effect modification by methylation capacity and dietary folate. Environmental research. 2018;164:124–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Desai G, Vahter M, Queirolo EI, Peregalli F, Mañay N, Millen AE, et al. Vitamin B-6 Intake Is Modestly Associated with Arsenic Methylation in Uruguayan Children with Low-Level Arsenic Exposure. The Journal of nutrition. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Carreón Valencia T, López Carrillo L, Romieu I. Manual de procedimiento en la toma de muestras biológicas y ambientales para determinar niveles de plomo. Manual de procedimiento en la toma de muestras biológicas y ambientales para determinar niveles de plomo1995. p. 85-. [Google Scholar]
- 41.Roy A, Queirolo E, Peregalli F, Manay N, Martinez G, Kordas K. Association of blood lead levels with urinary F(2)-8alpha isoprostane and 8-hydroxy-2-deoxy-guanosine concentrations in first-grade Uruguayan children. Environmental research. 2015;140:127–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Bradley RH, Caldwell BM, Corwyn RF. The Child Care HOME Inventories: Assessing the quality of family child care homes. Early Childhood Research Quarterly. 2003;18(3):294–309. [Google Scholar]
- 43.Barg G, Daleiro M, Queirolo E, Ravenscroft J, Mañay N, Peregalli F, et al. Association of Low Lead Levels with Behavioral Problems and Executive Function Deficits in Schoolers from Montevideo, Uruguay. International journal of environmental research and public health. 2018;15(12):2735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Wongsasuluk P, Chotpantarat S, Siriwong W, Robson M. Human biomarkers associated with low concentrations of arsenic (As) and lead (Pb) in groundwater in agricultural areas of Thailand. Scientific Reports. 2021;11(1):1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Kippler M, Skröder H, Rahman SM, Tofail F, Vahter M. Elevated childhood exposure to arsenic despite reduced drinking water concentrations—a longitudinal cohort study in rural Bangladesh. Environment international. 2016;86:119–25. [DOI] [PubMed] [Google Scholar]
- 46.Gibson JM, Fisher M, Clonch A, MacDonald JM, Cook PJ. Children drinking private well water have higher blood lead than those with city water. Proceedings of the National Academy of Sciences. 2020;117(29):16898–907. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Ahsan H, Chen Y, Parvez F, Argos M, Hussain AI, Momotaj H, et al. Health Effects of Arsenic Longitudinal Study (HEALS): description of a multidisciplinary epidemiologic investigation. Journal of exposure science & environmental epidemiology. 2006;16(2):191–205. [DOI] [PubMed] [Google Scholar]
- 48.Gilbert-Diamond D, Cottingham KL, Gruber JF, Punshon T, Sayarath V, Gandolfi AJ, et al. Rice consumption contributes to arsenic exposure in US women. Proceedings of the National Academy of Sciences. 2011;108(51):20656–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Lim C, Shaharuddin M, Sam W. Risk assessment of exposure to lead in tap water among residents of Seri Kembangan, Selangor state, Malaysia. Global Journal of Health Science. 2013;5(2):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Ryan PB, Huet N, MacIntosh DL. Longitudinal investigation of exposure to arsenic, cadmium, and lead in drinking water. Environmental Health Perspectives. 2000;108(8):731–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Medicine Io. Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate. Washington, DC: The National Academies Press; 2005. 638 p. [Google Scholar]
- 52.Madley-Dowd P, Hughes R, Tilling K, Heron J. The proportion of missing data should not be used to guide decisions on multiple imputation. Journal of clinical epidemiology. 2019;110:63–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Centers for Disease Control and Prevention C. Biomonitoring summary - lead. 2017.
- 54.National Research Council N. Arsenic in drinking water: National Academies Press; 1999. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.