Element Levels and Predictors of Exposure in the Hair of Ethiopian Children

Children’s development and health may be affected by toxic heavy metal exposure or suboptimal essential element intake. This study aimed to provide updated information regarding the concentrations of 41 elements in children’s hair (aged under 18) living in a rural area of the Benishangul-Gumuz region, Ethiopia. The highest average levels (as a geometric mean) for toxic heavy metals were obtained for Al (1 mg kg−1), Pb (3.1 mg kg−1), and Ni (1.2 mg kg−1), while the lowest concentrations among the essential elements were found for Co (0.32 mg kg−1), Mo (0.07 mg kg−1), Se (0.19 mg kg−1), and V (0.8 mg kg−1). Hair analysis was combined with a survey to evaluate relationships and variations among subgroups and potential metal exposure predictors. Females showed significantly higher concentrations for most hair elements, excluding Zn, than males, and the 6–11 years age group reported the highest levels for Be, Ce, Co, Fe, La, Li, Mo, and Na. The main predictors of exposure to toxic elements were fish consumption for Hg and drinking water for Ba, Be, Cs, Li, Ni, Tl, and U. The data from this study can be used to develop prevention strategies for children’s health and protection in developing countries.


Introduction
Elements present in the environment can come from natural (such as volcanic activity and forest fires) or anthropogenic (such as industrial, agricultural, and domestic) sources [1][2][3][4][5]. Factors that influence human exposure and, consequently, the presence of possible toxic effects are mainly dietary habits (i.e., fish, meat, cereals, vegetables, and water consumption) and outdoor and indoor air quality [6][7][8][9][10][11][12][13]. In particular, toxic heavy metals have no dietary value in humans [12,13]. More specifically, As, Hg, Pb, and Cd have been classified as the most dangerous elements affecting health [14]. As and Cd are classified as carcinogens (Group I), and Pb is ranked as a probable carcinogen (Group IIA) in humans by the International Agency for Research in Cancer (IARC) [15]. Instead, Hg, in organic [as methylmercury (MeHg)] or inorganic form, is classified by the IARC as a probable carcinogen in humans (Group IIB) or not classifiable as carcinogenic (Group III), respectively [15]. Instead, essential elements (such as Ca, Co, Cu, Fe, K, Mg, Mn, Mo, Na, P, Se, V, and Zn) are fundamental for the growth and health of children and perform several important functions in the human body, including bone formation, regulation of body fluids, and participation in the vital processes of cells [16,17]. However, The hair samples (n = 81) were taken from children aged under 18 (40 boys, 49.4%; 41 girls, 50.6%) whose hair had not been colored or treated. The study population's characteristics were collected using individual anonymous questionnaires with specific information on age, gender, weight, height, passive smoking, and dietary habits (water, meat, fish, vegetables, fruit, legumes, cereals consumption). The variables studied are shown in Table 1. Food consumption frequency by food groups was converted into quantitative intakes (g month −1 or mL month −1 ), according to other authors [57,58,68]. Monthly intake of individual foods was estimated according to the following formula: FI = W or V × IFr (1) where FI is the food intake (g month −1 or mL month −1 ), W = weight (g) or V = volume (mL) of the portion size, and IFr = intake frequency (number of portions) per month. This research was a non-interventional/observational study based on the definitions of the European Directive 2001/20/EC for which the approval of an Ethics Committee was not requested [69]; it was conducted according to the Helsinki Declaration (1964) and its later amendments and followed the International Code of Ethics for Occupational Health Professionals [70]. Collected information was used on aggregate health data of the children, with no possibility of individual identification. Before collecting the questionnaire information and the children's hair samples, all parents were informed about the study's aim and gave their consent. Children's hair sampling was performed at least with the presence of one of the parents.

Sample Collection
To avoid external contamination, hair samples (~0.05 g; ~1 cm long) were collected from the nape using stainless steel scissors and disposable vinyl gloves, as previously described [71,72]. Then, hair samples were stored in polyethylene bags until analysis at room temperature.
We also measured elements in the following drinking water: the Blue Nile river and the Blue Nile river's treated water. The water samples were collected at the same time as the hair samples in plastic urine collection cups, which were stored at −20 °C until analysis. The hair samples (n = 81) were taken from children aged under 18 (40 boys, 49.4%; 41 girls, 50.6%) whose hair had not been colored or treated. The study population's characteristics were collected using individual anonymous questionnaires with specific information on age, gender, weight, height, passive smoking, and dietary habits (water, meat, fish, vegetables, fruit, legumes, cereals consumption). The variables studied are shown in Table 1. Food consumption frequency by food groups was converted into quantitative intakes (g month −1 or mL month −1 ), according to other authors [57,58,68]. Monthly intake of individual foods was estimated according to the following formula: where FI is the food intake (g month −1 or mL month −1 ), W = weight (g) or V = volume (mL) of the portion size, and IFr = intake frequency (number of portions) per month. This research was a non-interventional/observational study based on the definitions of the European Directive 2001/20/EC for which the approval of an Ethics Committee was not requested [69]; it was conducted according to the Helsinki Declaration (1964) and its later amendments and followed the International Code of Ethics for Occupational Health Professionals [70]. Collected information was used on aggregate health data of the children, with no possibility of individual identification. Before collecting the questionnaire information and the children's hair samples, all parents were informed about the study's aim and gave their consent. Children's hair sampling was performed at least with the presence of one of the parents.

Sample Collection
To avoid external contamination, hair samples (~0.05 g;~1 cm long) were collected from the nape using stainless steel scissors and disposable vinyl gloves, as previously described [71,72]. Then, hair samples were stored in polyethylene bags until analysis at room temperature.
We also measured elements in the following drinking water: the Blue Nile river and the Blue Nile river's treated water. The water samples were collected at the same time as the hair samples in plastic urine collection cups, which were stored at −20 • C until analysis.

Hair Samples
Forty-one elements were analyzed, including essential of Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, P, Se, V, and Zn, and potentially toxic of Al, As, B, Ba, Be, Cd, Cs, Hg, Li, Ni, Pb, Sb, Sn, Tl, and U (see Table 1). Chemical analysis was performed in the Chemistry Laboratory of Sapienza University of Rome (Italy). Except for Hg, element concentrations in the hair were evaluated using an inductively coupled plasma mass spectrometer (ICP-MS; 820-MS Bruker, Bremen, Germany) equipped with a collision-reaction interface (CRI) and glass nebulizer (0.4 mL min −1 ). The data were collected according to a previously reported method [71]. Mercury was analyzed using a cold vapor generation atomic fluorescence spectrometer (CV-AFS; AFS 8220 Titan, FullTech Instruments, Rome, Italy), as described previously [72,73]. The ICP-MS analysis mode, measured isotopes, and internal standards are shown  in Supplementary Materials Table S1, while the preparation of both calibration and internal standards is described in Supplementary Materials Section 1.
Hair samples were digested using an analytical procedure described previously [71,73]. Briefly, 0.02 g hair samples were transferred into polypropylene tubes, mixed with 0.5 mL 67% HNO 3 (super-pure, Carlo Erba Reagents, Milan, Italy) and 0.25 mL 30% H 2 O 2 (super-pure, Merck KGaA, Darmstadt, Germany), and heated in a water bath system (WB12, Argo Lab Modena, Italy) for 20 min at 95 • C. The digest was left to cool down, and the contents of the tubes were diluted to 10 mL with deionized water (resistivity, ≤18.3 MΩ cm), filtered (0.45 µm pore size, GVS Filter Technology, Indianapolis, IN, USA), and analyzed by ICP-MS and, after further dilution in the ratio 1:1 with 6% HCl (assay >36%; Promochem, LGC Standards GmbH, Wesel, Germany), by CV-AFS. Each sample was analyzed in duplicate. Blanks were treated as samples for the subtraction of the background signal from the reagents.

Water Samples
Drinking water sampling and analyses were carried out following the Italian reference analytical methods [74,75]. Briefly, a volume of 250 mL (three replicates) of both treated and untreated water of the Blue Nile river was collected using a sterile polyethylene container with a screw cap. Each sample had an indelible identification label to identify it uniquely. Sampling was carried out where households obtained their drinking water. In particular, if there were more possibilities of access to the same water source (different access points to the Blue Nile river or different taps of the same treated water), a single sampling point was considered (Supplementary Materials Figures S2 and S3). After sampling, all water samples were transported to the laboratory using thermal bags and then filtered, acidified to 2% (v/v) HNO 3 (pH < 2), and stored at a temperature in the range 1-10 • C up to instrumental analysis. All water analyses were completed within one week after fieldwork. The water samples were analyzed diluted in a 1:10 and 1:100 ratio with 2% (v/v) HNO 3 and without dilution. Blanks and calibration standards were also made in 2% (v/v) HNO 3 . Internal standards (45Sc, 89Y, 103Rh, 115In, 232Th) were added to all samples, blanks, and calibration standards for ICP-MS analysis. To verify the accuracy, certified reference material (SRM 1643e trace elements in water; National Institute of Standards and Technology, NIST; Gaithersburg) was analyzed. The percent relative standard deviation for the repeatability did not exceed the 10% limit, and trueness bias percentages in the range −5 to 10% for all the studied elements were found.

Statistical Analysis
The statistical analysis was performed using SPSS version 25.0 for Windows (SPSS Inc., Chicago, IL, USA). Levels below DL were replaced by DL/2 [76]. Elements with a concentration <DL in more than 20% of the samples were excluded from the statistical calculation.
The most important descriptors such as arithmetic (AM) and geometric mean (GM), minimum and maximum levels, and the 25th, 50th, 75th, and 95th percentile were calculated to perform the descriptive statistical analysis.
The Shapiro-Wilk test and the Levene test were used to evaluate the normality and equal variances hypotheses, respectively [77]. All element concentrations were not normally distributed, so non-parametric methods such as the Mann-Whitney test and the Kruskal-Wallis test were used to compare element levels according to variable categories [78].
The relationship between the hair elements' concentrations and the factors considered was studied using the Spearman correlation test. For a value of p ≤ 0.05, a statistically significant relationship was considered.
A multiple linear regression model (backward method) was used to evaluate the influence of the study factors (gender, age, BMI, passive smoking, and water, meat, fish, fruit and vegetables, and cereals consumption) on the level of elements in children's hair. In each model, the natural log-element concentration was included as a dependent variable. The p value used as a criterion for the entry and stay of the variables was 0.05 and 0.1, respectively. The 95% confidence intervals were calculated for the model coefficients to evaluate the sample's statistical estimates [79,80]. The effect of potential outliers was checked using residual graphs.

Element Levels in Children's Hair
Descriptive statistics of element levels are given in Table 2. Elements with concentrations lower than the DL in more than 20% of the hair samples were excluded from the statistical analysis [Al (40.7%), As (35.8%), B (79.0%), Bi (44.4%), Nb (65.4%), Si (23.5%), Te (66.7%), and W (86.4%)]. The highest concentrations were found for the essential elements with Ca, Fe, K, Mg, and Na having a GM higher than 100 mg kg −1 , followed by P and Zn, which exhibited values of 98 and 86 mg kg −1 , respectively. Average concentrations (as GM) above 1 mg kg −1 were found, in order of abundance, for Si > Mn > Sr > Ti > Cu > Ba > Pb > Rb > B, and Ni, while element concentrations ≤ 1 mg kg −1 were detected for all other elements with decreasing contents in the following order: The possible exposure sources to the toxic elements analyzed may be several, such as contaminated food, water, and soil pollution, especially by the abuse or misuse of chemicals in agriculture, indoor and outdoor air pollution from inefficient solid fuel combustion (wood, animal dung, charcoal, crop wastes, and coal) and burning of waste, residues of manufacturing industrial products, mining, and oil refining [81][82][83]. As, Cd, Hg, and Pb are known to be of health concern in Africa, including in Ethiopia [81]. Figures 2-5 show a comparison of the mean levels found in this study for the priority and most toxic trace elements (As, Cd, Hg, and Pb, respectively) with those of other biomonitoring studies in children.
Water, food, air, and soil can be possible exposure sources to As from both natural geochemical processes (erosion) or anthropogenic pollution (arsenical insecticides, improper waste and sewage disposal, combustion of fossil fuels) [81]. For Cd (Figure 3), higher levels were obtained (GM: 0.1 mg kg −1 ) compared with data reported in children from all the reported studies, excluding those from Russia, in both exposed and unexposed populations (median: 0.11-0.12 mg kg −1 ) [87]. Exposure to Hg can occur mainly by the ingestion of contaminated fish (organic Hg, such as methylmercury) and small-scale artisanal gold mining (elemental Hg) [81]. Furthermore, gold extraction is routinely carried out near water sources, contaminating the environment and drinking water. The overall Hg concentrations ranged from 0.012 to 0.483 mg kg −1 with a GM of 0.056 mg kg −1 , which were the lowest compared with all data reported by other authors (Figure 4) [22,57,[85][86][87][88][89][90][91][92][93]. All the Hg hair data were also below the health-based values proposed (0.58, 1.0, and 2.3 mg kg −1 ) [59]. However, the Pb levels (GM: 3.1 mg kg −1 ; Figure 5) were the highest compared with all data reported in other studies [22,55,[84][85][86][87]94]. Hair can be used as a suitable biomarker of Pb exposure [95,96]. The reduction or removal of Pb from gasoline has produced a significant decline in pediatric morbidity. Llorente Ballesteros et al. [94] highlighted that the Pb content reduction in petrol was seen in environmental levels and, therefore, in human hair. However, children and adults continue to be exposed to Pb in Africa [81]. Here, most of the Pb in the environment comes from human activities such as waste combustion, mining, indiscriminate dumping, and even the use of Pb-based paints [81].
Some essential element average levels (Ca, Co, Cu, Mg, Mo, P, Se, V, and Zn) found in our study fell within the range of values usually reported for other children populations in Spain [94], Italy [55,97], and Russia [87]. However, the GM of children's hair Fe (243 mg kg −1 ), Mn (27 mg kg −1 ), and Na (1780 mg kg −1 ) obtained in our study was around one order of magnitude higher than that found in other studies worldwide [55,87,94,97].

Factor and Correlation Analysis
The population characteristics are shown in Table 1. A total number of 81 children's hair samples were collected, with a similar proportion of boys (51.1%) and girls (48.9%). Overall, 40.8% of children were aged 6-11 years and 29.6% were both <5 and 12-18-year-old. Of all children, 87.7% were underweight (BMI < 18.5).

Factor and Correlation Analysis
The population characteristics are shown in Table 1. A total number of 81 children's hair samples were collected, with a similar proportion of boys (51.1%) and girls (48.9%). Overall, 40.8% of children were aged 6-11 years and 29.6% were both <5 and 12-18-year-old. Of all children, 87.7% were underweight (BMI < 18.5).  [86], and in both exposed and unexposed populations (median: 0.02 and 0.03 mg kg −1 , respectively) [87]. Water, food, air, and soil can be possible exposure sources to As from both natural geochemical processes (erosion) or anthropogenic pollution (arsenical insecticides, improper waste and sewage disposal, combustion of fossil fuels) [81]. For Cd (Figure 3), higher levels were obtained (GM: 0.1 mg kg −1 ) compared with data reported in children from all the reported studies, excluding those from Russia, in both exposed and unexposed populations (median: 0.11-0.12 mg kg −1 ) [87]. Exposure to Hg can occur mainly by the ingestion of contaminated fish (organic Hg, such as methylmercury) and small-scale artisanal gold mining (elemental Hg) [81]. Furthermore, gold extraction is routinely carried out near water sources, contaminating the environment and drinking water. The overall Hg concentrations ranged from 0.012 to 0.483 mg kg −1 with a GM of 0.056 mg kg −1 , which were the lowest compared with all data reported by other authors (Figure 4) [22,57,[85][86][87][88][89][90][91][92][93]. All the Hg hair data were also below the health-based values proposed (0.58, 1.0, and 2.3 mg kg −1 ) [59]. However, the Pb levels (GM: 3.1 mg kg −1 ; Figure 5) were the highest compared with all data reported in other studies [22,55,[84][85][86][87]94]. Hair can be used as a suitable biomarker of Pb exposure [95,96]. The reduction or removal of Pb from gasoline has produced a significant decline in pediatric morbidity. Llorente Ballesteros et al. [94] highlighted that the Pb content reduction in petrol was seen in environmental levels and, therefore, in human hair. However, children and adults continue to be exposed to Pb in Africa [81]. Here, most of the Pb in the environment comes from human activities such as waste combustion, mining, indiscriminate dumping, and even the use of Pb-based paints [81].
Some essential element average levels (Ca, Co, Cu, Mg, Mo, P, Se, V, and Zn) found in our study fell within the range of values usually reported for other children populations in Spain [94], Italy [55,97], and Russia [87]. However, the GM of children's hair Fe (243 mg kg −1 ), Mn (27 mg kg −1 ), and Na (1780 mg kg −1 ) obtained in our study was around one order of magnitude higher than that found in other studies worldwide [55,87,94,97].

Factor and Correlation Analysis
The population characteristics are shown in Table 1. A total number of 81 children's hair samples were collected, with a similar proportion of boys (51.1%) and girls (48.9%). Overall, 40.8% of children were aged 6-11 years and 29.6% were both <5 and 12-18-year-old. Of all children, 87.7% were underweight (BMI < 18.5).

Gender and Age
The data obtained (Table 3, Supplementary Materials Table S2-S4, and Figure 6) show that both gender and age affect the levels of toxic elements in children.
chemicals in different ways [103]. Due to differences in kinetics, mode of action, or susceptibility, some toxic trace elements may affect the health of males and females differently [55]. In accordance with other authors [101,104], our results indicate that females may be exposed to toxic metals more than males. The results presented by Sanna et al. [96] suggest that hair is a reliable biomarker for determining population exposure levels to Pb pollution, and they indicate, in agreement with our study, that females tend to accumulate Pb in the hair more than males. The higher concentration of Sr and U in hair samples from females was also reported by other authors [101,105,106]. The high Sr, Pb, and U content in female hair can be considered a common gender characteristic due to the bone's possible release during the growth period, which generally occurs earlier and faster than in boys [101,107,108].
In the puberty period, adolescents need a high amount of Zn to maintain their skeletal growth, and this demand is generally more prevalent in girls [109]. Our results are in agreement with a previous study by De Prisco et al. [110] that showed a higher Zn content in boys' hair (244 ± 153 mg kg −1 ) than in that of girls (189 ± 34 mg kg −1 ). However, our data contrast with the results of several studies [50,55,85,96,97,[111][112][113][114] that found that female hair contains more Zn than males. The lower content of Zn in female hair compared to male hair may be related to the higher presence of Cd in females. Cd can replace Zn in many metal-enzymes, and Zn deficiency may also be a symptom of Cd toxicity [81]. Human exposure to Cd in Africa is thought to be mainly due to tobacco smoking and the consumption of contaminated vegetables and crops [81]. High levels of Cd contamination were reported in lettuce grown by irrigation using water from the Akaki River in Ethiopia [115].
In accordance with other authors [55,85,102,116], also Ni, considered a carcinogen [117] and responsible for the most common type of allergy [16], was found in higher concentrations in the hair of females than males. However, our results are in contrast with the data obtained by Barbieri et al. [113] and Peña-Fernández et al. [50], who showed that the hair of males contained more Ni than females (females = 0.23 ± 4.89 mg kg −1 vs. males = 0.3 ± 6.5 mg kg −1 ; and females = 0.38 ± 0.34 mg kg −1 vs. males = 0.58 ± 0.34 mg kg −1 , respectively). Figure 6. Influence of age on element levels (geometric mean, mg kg −1 ) in children's hair samples. A non-parametric Kruskal-Wallis test was applied. Different letters "a" and "b" for the same element show significant differences (p < 0.05). Table S2 and Figure  6. Concentrations of most elements (Be, Ce, Co, Fe, La, Li, Mo, and Na) in the 6-11 years age group were the highest and were significantly higher than those in the <5 years group. Element Figure 6. Influence of age on element levels (geometric mean, mg kg −1 ) in children's hair samples. A non-parametric Kruskal-Wallis test was applied. Different letters "a" and "b" for the same element show significant differences (p < 0.05). Gender-related differences were found for all elements excluding Ba, Ca, Cr, Cu, Hg, Sb, and Se, with higher concentrations found especially in females (Table 3). Element concentrations in the hair of females and males were particularly different for Be (five times higher in females than males) and for Cd, Ce, Co, Cs, Fe, Ga, K, La, V, and Zr (three times higher in females than in males). Only Zn was higher in male children than females. The current literature also reports gender-related hair element concentration differences [22,55,94,[98][99][100][101][102]. However, gender is often not adequately considered in the interpretation of hair analysis results [55]. Various factors (such as bodily growth, physiology, sexual hormones, and lifestyle) can affect boys' and girls' responses to exposure to chemicals in different ways [103]. Due to differences in kinetics, mode of action, or susceptibility, some toxic trace elements may affect the health of males and females differently [55]. In accordance with other authors [101,104], our results indicate that females may be exposed to toxic metals more than males. The results presented by Sanna et al. [96] suggest that hair is a reliable biomarker for determining population exposure levels to Pb pollution, and they indicate, in agreement with our study, that females tend to accumulate Pb in the hair more than males. The higher concentration of Sr and U in hair samples from females was also reported by other authors [101,105,106]. The high Sr, Pb, and U content in female hair can be considered a common gender characteristic due to the bone's possible release during the growth period, which generally occurs earlier and faster than in boys [101,107,108].

The age distribution of the elements is shown in Supplementary Materials
In the puberty period, adolescents need a high amount of Zn to maintain their skeletal growth, and this demand is generally more prevalent in girls [109]. Our results are in agreement with a previous study by De Prisco et al. [110] that showed a higher Zn content in boys' hair (244 ± 153 mg kg −1 ) than in that of girls (189 ± 34 mg kg −1 ). However, our data contrast with the results of several studies [50,55,85,96,97,[111][112][113][114] that found that female hair contains more Zn than males. The lower content of Zn in female hair compared to male hair may be related to the higher presence of Cd in females. Cd can replace Zn in many metal-enzymes, and Zn deficiency may also be a symptom of Cd toxicity [81]. Human exposure to Cd in Africa is thought to be mainly due to tobacco smoking and the consumption of contaminated vegetables and crops [81]. High levels of Cd contamination were reported in lettuce grown by irrigation using water from the Akaki River in Ethiopia [115].
In accordance with other authors [55,85,102,116], also Ni, considered a carcinogen [117] and responsible for the most common type of allergy [16], was found in higher concentrations in the hair of females than males. However, our results are in contrast with the data obtained by Barbieri et al. [113] and Peña-Fernández et al. [50], who showed that the hair of males contained more Ni than females (females = 0.23 ± 4.89 mg kg −1 vs. males = 0.3 ± 6.5 mg kg −1 ; and females = 0.38 ± 0.34 mg kg −1 vs. males = 0.58 ± 0.34 mg kg −1 , respectively).
The age distribution of the elements is shown in Supplementary Materials Table S2 and Figure 6. Concentrations of most elements (Be, Ce, Co, Fe, La, Li, Mo, and Na) in the 6-11 years age group were the highest and were significantly higher than those in the <5 years group. Element concentrations are reported according to gender and age group in Supplementary Materials Tables S3 and S4. Results showed significant differences in the hair of males with Be, Ce, Fe, La, and Ti levels in the age group of 6-11 years and Be, Mo, and U levels in the age group 12-18 years higher than those in the age group <5 years. A significant difference was observed for females with Na and Sb levels in the age group <5 years lower than those in the age group 6-11 and 12-18 years, and the age group 6-11 years, respectively.
The increase in the element concentrations in children's hair during the first years of childhood could be due to this group's physiological characteristics. Children drink more water, eat more food, and breathe more air per unit weight than adults and have higher absorption rates [94]. These findings are also in accordance with the study by Kordas et al. [118], who demonstrated that older age in children is associated with a lower risk of element exposure. A decrease in hair element content with age may be due to increased excretion of chemicals for organ development and maturation [86,119].

Passive Smoking
The hair of children with non-smoking parents contained higher Cu concentrations than that with smoking parents (cigarettes number ranging from 5 to 20), as shown in Supplementary Materials Table S5. Environmental tobacco smoke also appears to affect K and Na content, but not significantly. Other authors have observed higher levels of essential elements in the hair of non-smokers than in smokers [120,121]. This topic is very interesting and, in the future, it should be studied in-depth, perhaps considering a greater number of samples. Table 4 shows the element levels in the analyzed drinking water. Both the Blue Nile river's treated and untreated water show a low content of major and trace elements. However, the concentrations of many elements (Ba, Be, Ca, Ce, Co, Cr, Cs, Fe, Ga, Hg, K, La, Li, Mg, Mn, Mo, Na, Ni, P, Rb, Sr, Ti, Tl, U, V, and Zr) in the hair of children who drink water from the Blue Nile river are significantly higher. These results require more in-depth analysis and highlight the need for monitoring the water quality of the Blue Nile river with periodic sampling. Table 4. Element concentrations in drinking water (µg L −1 ) and in children's hair (mg kg −1 ) according to water consumption. The relation between independent variables (height, weight, BMI, and food consumption) and quantitative dependent variables (element concentrations) was also studied (Table 5). Considering the sociodemographic characteristics, a significant strong positive relationship (p < 0.01) with height, weight, and BMI was found in Cu [Spearman coefficient (r): 0.300, 0.327, and 0.397, respectively] and only with BMI in Sn (r: 0.315). A strong negative correlation was found between BMI and the following elements: Ca, Fe, Ga, P, Ti, and V. We analyzed the relationship between element levels and food group consumption, and we found a significant positive correlation (p < 0.01) between Ba and Hg levels and fish consumption (r: 0.320, and 0.427, respectively), and between Na content and cereals consumption (r: 0.470). A significant strong negative relationship (p < 0.01) with fish consumption was found in Zn (r: −0.342). No significant correlation with meat consumption was shown for all elements. It is known that fish is the major source of exposure to organic Hg [122]. The correlation between exposure to Hg and fish consumption is widely reported in the literature [57][58][59] and confirmed in the present study. Consistent with the results of the Spearman correlation test previously discussed, the children who consumed fish showed significant higher levels of Hg (GM = 0.078 mg kg −1 , p < 0.001) and Ba (GM = 14 mg kg −1 , p < 0.01) and lower content of Zn (GM = 76 mg kg −1 , p < 0.01) than the other children (GM = 0.034, 7, and 105 mg kg −1 for Hg, Ba, and Zn, respectively) (Supplementary Materials Table S6).

Predictors of Exposure
A multiple linear regression analysis was used to study the predictors of children's exposure to both toxic and essential trace elements (Table 6 and Supplementary Materials Table S7). Gender, age, BMI, passive smoking, drinking water, fish consumption, and fruit and vegetable consumption were significant predictors of exposure to several study population elements. The main predictor of Hg exposure in children's hair was fish consumption ( Table 6). The potential effect of age on Hg exposure is disputed, as several studies showed an increase in hair Hg with age [58,93,123,124], while many others did not observe any influence [125]. Consistent with the results of the Kruskal-Wallis test (Supplementary Materials Tables S2 and S3, and Figure 6), age is also a significant predictor of Mo and Na.
Drinking water appears to be the major exposure predictor of some toxic elements such as Ba, Be, Cs, Li, Ni, Tl, and U (Table 6) and essential and trace elements Ca, Ce, Co, Cr, Fe, Ga, La, Mg, Mn, Mo, P, Rb, Sr, Ti, V, and Zr (Supplementary Materials Table S7). Additional food and environmental monitoring are needed to determine the different exposure sources among children, causing differences in hair elements' concentrations. The data currently available on the elements found in drinking water do not justify these differences. According to the results of the Mann-Whitney analysis discussed above, gender is a significant predictor of Be, Cd, Li, and U levels (Table 6) and Ca, Ce, Co, Fe, Ga, K, La, Mn, Mo, NA, P, Rb, Sr, Ti, V, Zn, and Zr (Supplementary Materials Table S7).

Study Limitations
This study has some limitations. First of all, the group's size was small (81 children), and the present study is a cross-sectional study; therefore, it does not allow an evaluation over time.

Conclusions
Children represent a population that is particularly vulnerable to the developmental and neurotoxic effects of toxic elements and should be given special attention in biomonitoring programs.
We want to highlight the highest Pb concentrations in the hair of Ethiopian children studied compared to other literature data. A more in-depth study of this relevant finding is needed, combining different types of environmental monitoring and more extensive biomonitoring programs. The data obtained confirm gender and age as important key factors that must be taken into account in interpreting the hair analysis. The gender-based variations suggest that females are likely at greater risk for toxic element exposure than males. Dietary habits affect the elemental composition of hair in children. In particular, the hair of children who consumed fish and drank Blue Nile water contained higher Hg levels and other toxic elements (Ba, Be, Cs, Hg, Li, Ni, Tl, and U), respectively.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/17/22/8652/s1, Table S1: Isotopes, analysis mode, and internal standards, Table S2. Element concentrations in children's hair by age group, Table S3: Element levels (mg kg −1 ) in male hair according to age, Table S4. Element levels (mg kg −1 ) in male hair according to age, Table S5: Element levels (mg kg −1 ) in children's hair according to passive smoking, Table S6: Influence of fish consumption on element levels (mg kg −1 ) in children's hair, Table S7: Results of the backward multiple linear regression model analysis for major and trace elements, Figure S1: Landscapes of the studied area (Bameza in the Benishangul-Gumuz region along the Blue Nile in north-western Ethiopia, Africa), Figure S2: Sampling site of the Blue Nile river water (Bameza, Ethiopia), Figure S3: Sampling site of the treated water of the Blue Nile river (Bameza, Ethiopia).