ment 372 (2006) 49–57www.elsevier.com/locate/scitotenv

Science of the Total Environ

Heavy metal concentrations in the general population ofAndalusia, South of Spain

A comparison with the population within the area of influence ofAznalcóllar mine spill (SW Spain)

Fernando Gil a,⁎, Luis F. Capitán-Vallvey b, Esperanza De Santiago a, Julio Ballesta b,Antonio Pla a, Antonio F. Hernández a, Mario Gutiérrez-Bedmar c,

Joaquín Fernández-Crehuet c, Joaquín Gómez b, Olga López-Guarnido a,Lourdes Rodrigo a, Enrique Villanueva a

a Department of Legal Medicine and Toxicology. University of Granada, Spainb Department of Analytical Chemistry. University of Granada, Spain

c Department of Preventive Medicine and Public Health. University of Málaga, Spain

Received 29 November 2005; received in revised form 23 June 2006; accepted 6 August 2006Available online 12 September 2006

Abstract

Levels of metalloids (As – urine) and heavy metals (Hg – urine, Cd – whole blood and Zn – serum) were determined by atomicabsorption spectrometry in 601 subjects living in the area affected by the Aznalcóllar mine spill (SW, Spain) and compared with thoseof a representative sample (960 subjects) selected from the Andalusian community (non-affected area), southern Spain. Thecharacteristic parameters of the analytical method including uncertainty were determined for each metal. Potential associations ofmetal concentration with age, sex and body mass index as well as life-style habits (smoking, alcohol consumption and food habits)were explored. Concentrations of all the metals studied were statistically higher in the population of the affected area with respect tothat of the non-affected area in Andalusia, although levels were always lower or similar to the values reported for the generalpopulation and below occupational reference limits. In conclusion, there is a lack of evidence that the spill had any incidence onhuman health in the population living in the affected area. There are few references in scientific literature reporting values from largeseries of samples, and hence our data could be useful for further studies.© 2006 Elsevier B.V. All rights reserved.

Keywords: Heavy metal analysis; Atomic absorption spectrometry; Aznalcóllar; Arsenic (As); Mercury (Hg); Cadmium (Cd); Zinc (Zn); Mine spill;Environmental health

⁎ Corresponding author. Departamento de Medicina Legal yToxicología, Facultad de Medicina. Universidad de Granada. C/Avda. Madrid 11, 18071 Granada, Spain. Tel.: +34 958 24 35 46, +34958 24 99 30; fax: +34 958 24 61 07.

E-mail address: fgil@ugr.es (F. Gil).

0048-9697/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2006.08.004

1. Introduction

On 25 April 1998, part of a massive holding lagooncontaining pyrite ore processing waste failed and re-leased an estimated 5–6 million m3 of acidic and metal-rich sludge and water into the Agrio and GuadiamarRivers, located in the Aznalcóllar region (Andalusia,

50 F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

south west Spain), affecting 55 km2 (Grimalt et al.,1999; Simon et al., 1999; Vidal et al., 1999; Galán et al.,2002; Taggart et al., 2005). The Guadiamar river flowsthrough part of the Doñana National Park, one of themost important wildlife reserves in Europe, that is lo-cated north of the Guadalquivir River Estuary, in theprovince of Huelva. The main pollutants in the soilsafter spillage were Zn, Pb, As, Cu, Cd and Hg. Most ofthe Cu, Zn and Cd entered the soil as part of the solutionphase (Aguilar et al., 2004). The concentration ofwatersoluble metals (Zn, Cd and Cu) increased notablyand quickly after the spillage. Arsenic was a majorcomponent of the polluting sludge (Pain et al., 1998;Alastuey et al., 1999, Querol et al., 1999). The mainsource of this As may well have been arsenopyrite, asthe pyrite ore deposit mined at Aznalcóllar contained0.9% arsenopyrite (Almodóvar et al., 1998). Once theheavy metals enter the ecosystems, the biological com-munity including the human population can be affected.The study of the content of metals in different humanbody fluids (usually whole blood, urine or serum) can bea method to evaluate their effect on health (Gil and Pla,2001). Since the disaster a great number of studies havebeen performed on the levels of heavy metals in animals(oysters, clams, molluscs, crustaceans, fish, crayfish,birds, etc.), soils and sediments, etc. (Sánchez López etal., 2003; Aguilar et al., 2004; Solà et al., 2004; Riba etal., 2005a,b), although no information about humanexposure has been reported. These studies also reportthat past mining activity contributed to acid and metalcontamination of nearby rivers and therefore the mineinfluence areas appear to be more contaminated thanother areas in Andalusia.

Specific metal analyses in body fluids have beenassigned to a number of research groups within Spain.Thus, the determination of urinary As and Hg, blood Cdand serum Zn was assigned to the University of Granada(Department of Toxicology and Legal Medicine andDepartment of Analytical Chemistry).

The aim of the present study was to compare thelevels of selected heavy metals (As, Hg, Cd and Zn) inbody fluids from the population affected by the minespill in Aznalcóllar to those from a representative popu-lation of Andalusia used as reference (non-affectedarea). In addition, it should be pointed out that there arefew references in scientific literature reporting valuesfrom large groups of individuals, and hence the secondobjective of the study was to obtain reference values forthe general population that may be useful for compar-isons in future papers or when facing public healthproblems in which heavy metal contamination play arole.

The influence of other variables namely sex, age andbody mass index (BMI) as well as life-style habits in-cluding smoking, alcohol consumption and food habitson the metal concentrations was also studied. It is worthemphasizing that this paper is the first of its kind per-formed in Spain, as regards the Aznalcóllar mine spill.

2. Material and methods

2.1. Apparatus

Total As and Cd were determined by means of aPerkin-Elmer AAnalyst 800 Atomic Absorption Spec-trometry (Perkin Elmer, Norwalk, USA) equipped withZeeman background correction and an AS-800 auto-sampler. Arsenic was measured with direct flow injec-tion through hydride generation system (Perkin-ElmerFIAS-100) and cadmium by graphite furnace andgraphite tubes with integrated L'vov platform.

Zinc and mercury were determined in a Perkin ElmerLB1100 Atomic Absorption Spectrometry (PerkinElmer, Norwalk, USA) equipped with Power SupplyLamp System and MHS-10 Mercury Hydride System.

Urine creatinine was determined by the method ofJaffe using a Hitachi 917 autoanalyzer.

2.2. Reagents

Atomic absorption spectrometry standard solutionsfor As, Hg, Cd and Zn (Titrisol grades fromMerck) wereused to build up the calibration curve. They were pre-pared from a stock solution of 1000 mg/L for each metalby successive dilutions with water. All aqueous solutionsof reagents and standards were prepared using reverse-osmosis type quality water produced by a Milli-RO 12plus Milli-Q purification system (Millipore, Bedford,MA).

The chemicals used were all of analytical reagentgrade. High-quality concentrated (65% w/v) nitric acid(Merck, Darmstadt, Germany), (96%w/v) sulphuric acid(Merck) and (37% w/v) hydrochloric acid (Merck), so-dium borohydride (Merck), sodium hydroxide (Panreac,Barcelona, Spain), ascorbic acid (Panreac), potassiumiodide (Panreac), magnesium nitrate (Merck), palladiumnitrate (Merck), Triton X-100 (Merck), potassium per-manganate (Merck) and silicone antifoaming agent(Merck) were used.

Volumetric polyethylene material (including auto-sampler cups) and micropipettes with plastic tips wereused. The glass material was cleaned by soaking in 20%v/v HNO3 for 24 h. It was finally rinsed with Milli-Q®water and dried in a polypropylene container.

51F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

2.3. Validation of analytical methods

The characteristic parameters of the analyticalmethodswere determined for each metal (As, Hg, Cd and Zn) bymeans of analysis of blanks and standard solutions atdifferent concentrations. These parameters included thelimit of detection (LOD) and quantification (LOQ) asdetermined following the recommendations of the IUPAC(Sánchez López et al., 2003), linear range, precision(minimal, intermediate and reproducibility), accuracy,recovery and characteristic mass. Uncertainty of methodswas calculated for each metal. The overall uncertaintymust be estimated considering every source of uncertaintyand treating it separately to obtain the contribution fromthat source. Each of the separate contributions to uncer-taintymay be referred to as an uncertainty component. It isoften possible to evaluate the combined effect of severalcomponents as a global standard uncertainty (Eurachem/Citac Guide, 2000). Once the parameters and their asso-ciated uncertainties that contribute to the uncertainty forthe method as a whole are listed, the individual uncer-tainties are combined in the uncertainty budget in whichwe include the relative reference standard uncertainty andthe relative balance calibration standard uncertainty (gra-vimetric steps) and the volumetric sample relative uncer-tainty (volumetric steps). Statistical evaluation of therelative uncertainty associated with recovery involves therelative uncertainty associated with the calibration curve(including addition method in the particular case of As),the relative uncertainty associatedwith the reproducibilityof the method and statistical evaluation of the relativeuncertainty associated with the intermediate precision ofthe method (quantification step).

2.4. Population study and biological samples

A total number of 1561 individuals were included inthe present study, 601 of them were from the areaaffected by the toxic spill (Aznalcázar, Aznalcóllar,Benacazón, Carrión de los Céspedes, Huévar, Pilas,Sanlúcar la Mayor, Villamanrique de la Condesa andHinojos) and 960 from representative Andalusian areasnot affected by the spillage. A minimum sample size of500 per group was calculated in order to achieve enoughstatistical power to detect differences below 10%with anα error of 0.05 and a β error of 0.1. A random doublesampling was carried out. In the first stage 50Andalusian towns were selected and in the second onea representative sample of individuals was randomlyselected from the 1996 census (according to gender andthree age categories: 12–24, 25–59, and 60–75 yearolds).

Different types of biological samples were analyzedin this study, urine, whole blood and serum where Asand Hg, Cd and Zn were measured, respectively.

A previously validated food frequency questionnaire(Gómez Aracena, 1990) was given to all participants inorder to provide the following information: sex, age,BMI and life-style habits (smoking, alcohol consump-tion and food habits). The two population samples(affected and non-affected) can be taken as comparablein health, life-style and living conditions. None of thesubjects reported occupational exposure to any of themetal elements determined in this study.

2.5. Analytical procedures

Metal concentrations in urine were adjusted forcreatinine levels.

2.5.1. ArsenicA direct flow-injection atomic absorption spectro-

metric technique (FI-HGAAS) was used to measureurinary levels of total arsenic. The arsenic contained instandard solutions (calibration curve 0, 0.5, 1.5 and2.5 μg/L) or urine samples was reduced to As3+ prior toanalysis with a mixture of potassium iodide and as-corbic acid. To 1 mL of sample or reference solution1 mL of concentrated HCl and 1 mL of 5% (w/v) KI –ascorbic acid was added. After 45 min at room tem-perature the mixture was diluted to 10 mL with water.The reducing agent was an aqueous solution of 0.2%(w/v) NaBH4 in a 0.05% (w/v) NaOH solution freshlyprepared and filtered. Standard addition was required.The parameters for As were wavelength 193.7 nm,integration time 15 s, smoothing 19 points or 0.5 s andtemperature cell of 900 °C. An electrodeless dischargelamp was used.

2.5.2. MercuryMercurywas determined using an aqueous solution of

3% (w/v) NaBH4 in a 1% (w/v) NaOH solution freshlyprepared and filtered as reducing agent. One to two dropsof silicone antifoaming was dispensed into a reactionflask before introducing any solution. All solutions werestabilized in the reaction flask by adding 250 μL of 5%KMnO4 solution before starting the determination. Themercury standard calibration plot (0, 20, 50, 100, 200 μg/L) was prepared in 10 mL of acid mixture containing1.5% HNO3 and 1.5% H2SO4. 9 mL of acid mixture wasadded to 1 mL of urine sample. The parameters for Hgwere wavelength 253.6 nm with time delay of 12 s. Noflame was required and an electrodeless discharge lampwas used.

52 F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

2.5.3. CadmiumA calibration curve with different cadmium concen-

trations (0, 1, 3, 5 μg/L) was prepared in 0.2% HNO3.Aliquots of 20 μL of whole blood were introduceddirectly into the graphite furnace with an equal volumeof matrix modifier (a mixture of 3.3% Pd and 0.03%Mgas nitrates in 0.2% HNO3). The temperature programwas previously optimized. Two pyrolisis steps wererequired at 600 and 700 °C. Atomization temperaturewas set at 1600 °C with 0 mL/min argon flow rate. Anelectrodeless discharge lamp was used and the wave-length set at 228.8 nm.

2.5.4. ZincCalibration standards (0, 50, 100, 200, 400 mg/L)

were prepared in 5% glycerol solution to match serumviscosity. Serum samples were diluted 1:5 by using thesame solution. Operational conditions were a wave-length of 213.9 nm with a time delay of 1 s. A hollowcathode lamp was used.

2.6. Reference materials

Reference samples for whole blood (three levels, ref.201505, 201605 and 201705), urine (ref. 201205) andserum (ref. 201405) were supplied by Seronorm (Bill-ingstad, Norway). As they were supplied freeze-dried,they were reconstituted by adding 5 mL of water.

2.7. Statistical analysis

Data were analysed by using the software packageSPSS 11.0 (SPSS, Chicago, USA). Measurements belowthe detection limit were replaced by LOD/2. Metal

Table 1Summary of results for the characteristic parameters of the analytical methodthe general population of Andalusia and that of the area of influence of Azn

As

LOD/LOQ (μg/L) 0.03/0.10Linear range (μg/L) 4Linear equation y=0.0800x+0.0021r 0.9998RSD (%) (n=10) 0.0008Precision (%) Minimal a 3.51

Intermediate b 6.70Reproducibility a 5.09

Accuracy (%) 1.33Recovery (%) 101.04Characteristic mass (pg) 25.37Uncertainty (%) 5.10

LOD: limit of detection; LOQ: limit of quantification; RSD: standard relativa n=10.b The analyses were carried out in 1-week periods for 5 weeks.

concentrations in urine were adjusted for creatinine con-centration and regressions were performed on creatinine-corrected data. Log-transformed metal concentrationswere used to normalize their distributions.

The statistical power to detect differences in the heavymetal levels of the affected population as compared to thenon-affected population was 79.6% for arsenic, 56.6%for mercury, 89.3% for cadmium and 89.6% for zinc.

The potential influence of classical confounders(tobacco, alcohol and BMI) on heavy metal levels wasassessed by Student's t test in non-affected populationwhich was considered as control. The association bet-ween levels of heavy metals and BMI was assessed bySpearman's rank correlation test.

Differences in the mean heavy metal levels betweenthe affected and non-affected populations were assessedby the analysis of covariance (ANCOVA) after adjustingfor age, gender and any other confounder which pre-viously showed any association in the bivariate analysis.Since the effect of each covariate may not be the same forthe two populations studied, the interaction effectbetween each covariate and the population was checked.

3. Results and discussion

The characteristic parameters of the analytical me-thod for each metal are presented in Table 1. In all casesthe correlation coefficients were higher than 0.99 and theresults showed that the instrumental response can beconsidered linear in the range studied. The standardaddition method was applied to calculate the recovery ofeach metal.

Table 2 shows the influence of smoking, alcoholconsumption and sex on geometric mean levels of the

for the determination of urinary As and Hg, blood Cd and serum Zn inalcóllar mine spill

Hg Cd Zn

0.002/0.007 0.03/0.09 26.06/86.8620 7 400y=1.5941x+0.0083 y=0.0430x+0.0035 y=0.0002x+0.00290.9992 0.9996 0.99820.0802 0.0004 0.08931.98 2.62 1.833.56 4.94 4.532.20 3.45 8.215.80 3.34 3.06104.83 100.50 103.3112.00 2.96 7271.2011.49 2.71 12.28

e deviation.

Table 2Influence of smoking and alcohol drinking habits and sex, age and body mass index (BMI) on geometric mean levels of urinary As and Hg (μg/gcreatinine), blood Cd (μg/L) and serum Zn (μg/L) in the non-affected population

Smoking habits P Alcohol P Sex P Spearman's correlation

Yes No Yes No Male Female Age (P) BMI (P)

As 1.20 1.34 0.44 1.21 1.38 0.33 1.36 1.22 0.42 0.08 (0.07) 0.003 (0.946)Hg 1.14 1.17 0.87 1.18 1.13 0.68 1.04 1.28 0.05 0.15 (b0.001) 0.107 (0.013)Cd 0.46 0.08 b0.001 0.38 0.12 0.002 0.13 0.15 0.35 0.06 (0.18) 0.001 (0.981)Zn 1015.75 963.69 0.11 992.19 967.92 0.42 1021.25 943.19 0.009 −0.04 (0.37) −0.027 (0.514)

P: statistical significance.

53F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

metals studied (As, Hg, Cd and Zn) in the non-affectedpopulation as well as the correlation between age andBMI with those metal levels. Blood Cd levels weresignificantly higher in current smokers (6-fold withregard to non-smokers) and alcoholconsumers (3-fold).Our data are in accordance with previous studies thatreported an increased Cd level in whole blood fromsmokers with respect to non-smokers (Graasmick et al.,1985; Schumacher et al., 1993; Benedetti et al., 1994;Chia et al., 1994; Dell'Omo et al., 1999; Moreno et al.,1999). Urinary Hg levels showed a significant correla-

Table 3Geometric mean levels of urinary As and Hg (μg/g creatinine), blood Cd (μg(non-affected area) and the affected area of Aznalcóllar mine spill

Element Non-affected(95% CI)

Affected(95% CI)

Statisticalsignificance

As(a,b) 1.29(1.15–1.46)

1.73(1.47–2.04)

0.005

Hg(a,b,c) 1.16(1.05–1.29)

1.40(1.22–1.61)

0.034

Cd(a,b,d,e) 0.14(0.13–0.16)

0.19(0.16–0.22)

0.001

Zn(a,b,d) 979.67(954.50–1005.48)

1051.73(1016.08–1088.64)

0.001

Adjusted for: (a) age, (b) sex, (c) body mass index, (d) tobacco, and (e) alco1Lauwerys and Hoet, 1993; 2Concha Quezada, 2001; 3Baselt, 1980; 4Khordi-M1990; 8Staessen et al., 1990; 9Barany et al., 2002; 10De Mateo Silleras et al.⁎ As (μg/L) in non affected area was 2.19 (CI 95%: 2.05–2.33) and affec⁎⁎ Hg (μg/L) in non affected area was 2.01 (95% CI: 1.82–2.20) and affec

tion to age, so that older people presented increased Hglevels. Likewise, those that had greater BMI showedhigher urinary Hg levels. These two findings support theaccumulative potential of this metal. Moreno et al.(1999) found that age strongly influenced serum Znlevels but made only a small contribution to blood Cdconcentrations. Regarding sex differences, Zn concen-tration was significantly higher in Andalusian males thanfemales and the reverse was true for Hg levels.

Table 3 shows geometric mean levels of As, Hg, Cdand Zn adjusted for age, gender, BMI, tobacco and

/L) and serum Zn (μg/L) adjusted for several confounders in Andalusia

(P)Occupationalreference value

Other reference values

b40 μg/g creatinine1 b10 μg/L2, ⁎

Non-exposed: 10–300 μg/L3

After fish intake: 200–1700 μg/L3

b5 μg/g creatinine1 Non-exposed population:b10 μg/L3, ⁎⁎

Non-exposed workers:119 μg/L3, ⁎⁎

Exposed workers (asymptomatic):403 μg/L3, ⁎⁎

Children: 3.83±2.45 μg/L4, ⁎⁎

(n=43)5 μg/L3 Japan female: 1.58–3.82 μg/L5

(n=371)Healthy adults: 1–4 μg/L3

Non smokers: 1 μg/L6 (adults)0.1–1.7 μg/L7

Non smokers: 0.7 μg/L8

b1700 μg/L1 Healthy donors: 660–1020 μg/L3

(n=17)Swedish adolescents: 990 μg/L9

(n=372)Spanish donors: 972.2 μg/L10

(n=186)

hol.ood et al., 2001; 5Ikeda et al., 1997; 6Ewers et al., 1999; 7Minoïa et al.,

, 2000.ted area was 2.35 (CI 95%: 2.16–2.54).ted area was 2.13 (95% CI: 1.97–2.30).

54 F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

alcohol in our target populations (affected and non-affected by the mine spill). The affected populationshowed significantly higher levels of all the metalsstudied with respect to the non-affected population.Nevertheless, levels fall within their respective referenceranges. Reference values from healthy population as wellas exposed and non-exposed workers are also shown inTable 3. For comparison purposes, values from generalpopulation are more appropriate, nevertheless we havealso included occupational reference values because ofthe scarcity of existing data for the general population.The latter are not often comparable owing to a lack ofhomogeneity, as there are differences in units, bodyfluids, sample size, life-style influences (dietary intake,smoking habits) and even age.

For every metal, the interaction effect between popu-lation from the affected or non-affected area and all thecovariates listed in Table 2 was checked. However, nosignificant association was found for any of the inter-action terms assessed.

The mean urinary Hg levels were also significantlyhigher in the affected population, although they aresimilar to values reported for the general population(Ewers et al., 1999) and 4-fold lower when compared tooccupational reference values [b5 μg/g creatinine](Lauwerys and Hoet, 1993). The population of theaffected area presented higher blood Cd levels. Never-theless, the mean levels were below the reportedreference values (Baselt, 1980; Minoïa et al., 1990;Staessen et al., 1990; Ikeda et al., 1997; Ewers et al.,1999). The higher levels of mercury and cadmium in thepopulation affected by the mine spill might be due to thecumulative potential of these metals. Both of them weretwo of the main soil pollutants after the spillage, so thattheir long biological halflife and risk of accumulation inthe body could account for the differences found in thestudy. Anyway these differences should not raise any

Table 4Geometric mean levels – adjusted for the potential confounders (see footnoteZn (μg/L) in Andalusian Population (excluding the affected area)

Province As(a,b) Hg(a,b,c)

Average (95% CI) n Average (95% CI) n

Almería 1.34 (0.84–2.13) 40 2.81 (1.96–4.01) 38Cádiz 1.52 (1.12–2.05) 96 0.73 (0.58–0.91) 95Córdoba 3.20 (1.94–5.28) 35 0.84 (0.60–1.22) 35Granada 2.10 (1.47–3.01) 69 1.23 (0.94–1.62) 65Huelva 0.96 (0.65–1.41) 58 0.62 (0.46–0.82) 56Jaén 0.41 (0.28–0.60) 57 1.38 (1.03–1.85) 57Málaga 1.57 (1.22–2.02) 140 1.08 (0.89–1.30) 136Sevilla 0.81 (0.55–1.20) 57 2.86 (2.13–3.83) 56

Adjusted for: (a) age, (b) sex, (c) body mass index, (d) tobacco, and (e) alco

concern for public health as their Hg and Cd levels fallwithin reference values for the general population.

Serum Zn levels were significantly higher in theaffected population but similar to those reported for thegeneral population (Baselt, 1980; De Mateo Silleras etal., 2000; Barany et al., 2002) and 50% lower than theoccupational reference level (Lauwerys and Hoet,1993). These differences cannot be attributed to dietarypatterns since Zn levels failed to be associated with anyof the dietary items listed in Table 5. An alternativeexplanation can be found in background contamination,since Zn levels from the general population of Sevilleprovince were about 20% higher than those found in theaffected population (see later) which also belongs to theprovince of Seville (the province where the spillage tookplace).

Urine concentrations of As were statistically higherin the affected area with respect to the non-affected area.Nevertheless, levels were lower than reported values forgeneral population (Baselt, 1980; Concha Quezada,2001) and 20-fold lower than occupational referencelevels (Lauwerys and Hoet, 1993). We do not have asatisfactory explanation for these differences. AlthoughAs is strongly related to dietary items (Table 5), nodifferences exist in the dietary pattern between theaffected population and the remaining non-affectedpopulation from Seville. Environmental contaminationby As can also be discarded since arsenic levels fromthe non-affected population of the province of Sevilleare roughly half those found in the affected population.We could also speculate with industrial pollution, butthe main province with pyrite mines is Huelva andarsenic levels of their population were also lower thanthose from the affected area. The only explanation thatmight support our findings are differences in the arseniclevel in drinking water, although there is no evidence torule out or confirm this possibility. No known water

) – of urinary As and Hg (μg/g creatinine), blood Cd (μg/L) and serum

Cd(a,b,d,e) Zn(a,b,d)

Average (95% CI) n Average (95% CI) n

0.10 (0.07–0.15) 42 886.84 (809.44–971.64) 420.09 (0.07–0.12) 73 790.29 (744.68–838.69) 990.08 (0.06–0.12) 35 1087.27 (983.78–1201.64) 350.26 (0.20–0.35) 70 1183.48 (1101.52–1271.53) 680.17 (0.13–0.23) 58 795.44 (734.42–861.54) 550.13 (0.09–0.17) 59 711.22 (658.46–768.21) 590.17 (0.14–0.21) 171 1129.43 (1077.84–1183.50) 1600.10 (0.08–0.14) 59 1292.80 (1197.00–1396.26) 59

hol.

Table 5Spearman correlation between urinary As and Hg, blood Cd and serumZn levels and frequency of several foods intake evaluated by foodquestionnaire in subjects from the non-affected area (n=551 to 566)

As Hg Cd Zn

Eggs −0.095 ⁎ −0.128 ⁎⁎Mixed salad 0.105 ⁎

Pork −0.138 ⁎⁎Beef −0.087 ⁎Chicken −0.105 ⁎Pink Ling 0.206 ⁎⁎

Hake 0.103 ⁎

Anchovy 0.090 ⁎

Sardine 0.118 ⁎⁎ 0.094 ⁎

Prawn 0.104 ⁎

Shrimp 0.124 ⁎⁎

Crab 0.109 ⁎⁎

Mussel 0.110 ⁎⁎

⁎ Significant correlation at 0.05 level (bilateral).⁎⁎ Significant correlation at 0.01 level (bilateral).

55F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

supply with typically higher levels of arsenic exists inAndalusia.

Previous studies have reported high metal concentra-tions in the Guadiamar River in the late 1970s, wellbefore the accident, due very likely to past mining acti-vities when this river became one of the world's mostmetalpolluted (Solà et al., 2004). Riba et al. (2005a)have reported that the upper river Guadiamar still hassome traces of toxic mud from the mining spill.

Since 1999 the metal contamination caused by thespill is undergoing both fluvial and aerial redistributionand a continued input from the heavily contaminatedGuadiamar Valley/River may still be occurring and maycontinue to do so for many years (Taggart et al., 2005,2006). As a result, potential transfer of pollutants frommetal-accumulating macrophytes to herbivores mayoccur although it still remains to be studied (Taggart etal., 2005). This may be a significant food chain transferpathway which could possibly reach the human popu-lation close to the affected area. Aerial redistributionmay also have contributed to this environmentalcontamination. Both routes of exposure might explainthe higher metal levels in the population affected by thespill.

Table 4 shows mean levels of As, Hg, Cd and Zn inthe eight Andalusian provinces. Differences in dietaryhabits among these provinces (especially as regards tothe consumption of fish and seafood, which presenthigher As content) may have contributed to the diffe-rences observed and could account for the variability ofAs levels among the provinces. Interestingly, theaverage As levels in the affected area (Table 3) weredouble those found in the remaining province of Seville(Table 4), which seems to pinpoint a backgroundcontamination in the former area, although these levelswere of no public health concern as they were lower thaneither general population or occupational referencevalues (see Table 3). Urinary Hg also showed a highvariability, because their levels were significantly higherin the provinces of Seville and Almería. Hg concentra-tion from the affected area was roughly half that from theremaining province of Seville. Regarding serum Zn,lower levels were also observed in the affectedpopulation when compared to the rest of the Sevillepopulation.

Table 5 summarizes the correlation between metallevels (As, Hg, Cd and Zn) and food intake evaluated bya food questionnaire. Data correspond to the populationfrom the non-affected area. The ingestion of sardines andmussels exerted a strong influence on urine As levels.However the intake of meat (pork) was inversely relatedto As concentration in urine. A weaker association was

also observed with the intake of eggs, chicken, hake andanchovies. These findings suggest that the intake ofanimal products (chicken, pork, and eggs) is usuallyassociated with lower levels of arsenic in urine whichmay be related to less fish consumption. A significantassociation was found between whole blood Cd levelsand fish, seafood and mixed salad intake, which wasextremely notable in the case of pink ling – Genypterusblocades (r=0.206; Pb0.001). In turn, an inverse corre-lation between urinary Hg and eggs and beef intake wasobserved. Finally, serum Zn levels failed to be associatedwith food intake.

Juresa and Blanusa (2003) found the highestconcentrations of Hg and As in hake (Merlucciusmerluccius; 0.37 and 23.30 mg/kg fresh weight,respectively) and cadmium levels were about 10 foldhigher in shellfish (Mytilus galloprovincialis; 0.14 mg/kg fresh weight) than in fish from the Adriatic Sea.However, the estimated dose of those trace elementsfrom sea food consumption by the general populationdid not exceed the provisional weekly intake recom-mended by the Joint FAO/WHO Expert Committee onFood Additives. Llobet et al. (2003) found that theconsumption of fish (hake and sardine) and shellfish(mussel) was the main food item responsible for As,Hg and Cd intakes.

In conclusion, the metal concentrations of thepopulations studied from both affected and non-affectedareas by the mine spill were in all cases below referencevalues. Therefore, there is no evidence that the spill hadany impact on the population from the affected area andthat remediation measures carried out after theAznalcóllar pyrite mine spill could have contributed

56 F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

to the reduction of the environmental contaminationcapable of affecting the human population.

Acknowledgements

The authors would like to acknowledge the financialsupport given by Consejería de Medio Ambiente(Contrato de investigación no. 1581) and Consejeríade Salud (Contrato de investigación no. 2098) of Juntade Andalucía, Spain. The authors wish to thank Isabel J.Macdonald for her assistance in the language reviewing.

References

Aguilar J, Dorronsoro C, Fernández E, Fernández J, García I, Martín F,et al. Soil pollution by a pyrite mine spill in Spain: evolution intime. Environ Pollut 2004;132:395–401.

Alastuey A, García-Sánchez A, López F, Querol X. Evolution of pyriteand weathering and mobility of heavy metals in the Guadiamarvalley after the Aznalcóllar spill, SW Spain. Sci Total Environ1999;242:41–55.

Almodóvar GR, Sáez R, Pons JM, Maestre A, Toscano M, Pascual E.Geology and genesis of the Aznalcóllar massive sulphidedeposits , Iberian Pyri te Belt , Spain. Miner Depos1998;33:111–36.

Barany E, Bergdhal IA, Bratteby LE, Lundh T, Samuelson G, SchutzA, et al. Trace elements in whole blood and serum from Swedishadolescents. Sci Total Environ 2002;286(1–3):129–41.

Baselt RC. Biological monitoring methods for industrial chemicals.Davis, California: Biomedical Publications; 1980.

Benedetti JL, Devaily E, Turcotte F, Lefebvre M. Unusually highblood cadmium associated with cigarette smoking among threesubgroups of the general population, Quebec, Canada. Sci TotalEnviron 1994;152:161–7.

Chia SE, Chan OY, Sam CT, Heng BH. Blood cadmium levels in non-occupationally exposed adult subjects in Singapore. Sci TotalEnviron 1994;145:119–23.

Concha Quezada G. Metabolism of inorganic arsenic and biomarkersof exposure. Doctoral Thesis. Institute of Environmental Medicine.Stockolm, 2001.

Dell'OmoM, Muzi G, Piccinini R, Gambelunghe A, Morucci P, FiordiT, et al. Blood cadmium concentrations in the general populationof Umbria, central Italy. Sci Total Environ 1999;226:57–64.

De Mateo Silleras B, Pérez García A, Miján de la Torre A. The zincstatus in a selected Spanish population. A multivariate analysis.Nutr Hosp 2000;15:32–41.

Eurachem/Citac Guide. Quantifying uncertainty in analytical measure-ments 2nd edition. 2000. http://eurachem.bam.de/guides/quam2.pdf.

Ewers U, Krause C, Schulz C, Wilhelm M. Reference values andhuman biological monitoring values for environmental toxins. IntArch Occup Environ Health 1999;72:255–60.

Galán E, González I, Fernández-Caliani JC. Residual pollution load ofsoils impacted by the Aznalcóllar (Spain) mining spill after clean-up operations. Sci Total Environ 2002;286:167–79.

Gil F, Pla A. Biomarkers as biological indicators of xenobioticexposure. (Review Article) J Appl Toxicol 2001;21:245–55.

Gómez Aracena J. La encuesta alimentaria como instrumento deevaluación del estado nutritivo de la comunidad. Tesis Doctoral.Universidad de Málaga. 1990.

Graasmick C, Huel G, Moreau T, Sarmini H. The combined effect oftobacco and alcohol consumption on the level of lead and cadmiumin blood. Sci Total Environ 1985;41:207–17.

Grimalt JO, Ferrer M, Macpherson E. The mine tailing accident inAznalcóllar. Sci Total Environ 1999;242:3-11.

Ikeda M, Watanabe T, Zhang ZW, Moon CS, Shimbo S. The integrityof the liver among people environmentally exposed to cadmium atvarious levels. Int Arch Occup Environ Health 1997;69:379–85.

Juresa D, Blanusa M. Mercury, arsenic, lead and cadmium in fish andshellfish from the Adriatic Sea. Food Addit Contam2003;20:241–6.

Khordi-MoodM, Sarraf-Shirazi AR, Balali-MoodM. Urinary mercuryexcretion following amalgam fillig in children. J Toxicol ClinToxicol 2001;39:701–5.

Lauwerys R, Hoet P. Industrial chemical exposure. Guidelines forbiological monitoring. 2nd ed. Boca Raton: Lewis Publishers;1993.

Llobet JM, Falcó G, Casas C, Teixidó A, Domingo JL. Concentrationsof arsenic, cadmium, mercury and lead in common foods andestimated daily intake by children, adolescents, adults, and seniorsof Catalonia, Spain. J Agric Food Chem 2003;51:838–42.

Minoïa C, Sabbioni E, Apostoli P, Pietra R, Pozzoli L, Gallorini M,et al. Trace element reference values in tissues from inhabitantsof European Community: I. A study of 46 elements in urine,blood and serum of Italian subjects. Sci Tot Environ1990;95:89.

Moreno MA,Marin C, Vinagre F, Ostapczuk P. Trace element levels inwhole blood samples from residents of the city Badajoz, Spain. SciTotal Environ 1999;229:209–15.

Pain DJ, Sánchez A, Meharg AA. The Doñana ecological disaster:contamination of a world heritage estuarine marsh ecosystem withacidified pyrite mine waste. Sci Total Environ 1998;222:45–54.

Querol X, Alastuey A, López-Soler A, Plana F, Mesas A, Ortiz L, et al.Physico-chemical characterization of atmospheric aerosols in a ruralarea affected by the Aznalcollar toxic spill, south-west Spain duringthe soil reclamation activities. Sci Total Environ 1999;242:89-104.

Riba I, Blasco J, Jiménez-Tenorio N, González de Canales ML, AngelDelvalls T. Heavy metal bioavailability and effects: I. Bioaccu-mulation caused by mining activities in the Gulf of Cádiz (SW,Spain). Chemosphere 2005a;58:659–69.

Riba I, Blasco J, Jiménez-Tenorio N, González de Canales ML, AngelDelvalls T. Heavy metal bioavailability and effects: II. Histopathol-ogy–bioaccumulation relationships caused bymining activities in theGulf of Cádiz (SW, Spain). Chemosphere 2005b;58:671–82.

Sánchez López FJ, Gil García MD, Sánchez Morito NP, MartínezVidal JL. Determination of heavy metals in crayfish by ICP–MSwith a microwave-assisted digestion treatment. Ecotoxicol EnvironSaf 2003;54:223–8.

Schumacher M, Domingo JL, Llobet JM, Corbella J. Variability ofblood lead levels in urban population in relation to drinking andsmoking habits. Sci Total Environ 1993;138:23–9.

Simon M, Ortiz I, García I, Fernandez E, Fernandez J, Dorronsoro C,et al. Pollution of soils by the toxic spill of a pyrite mine(Aznalcóllar, Spain). Sci Total Environ 1999;242:105–15.

Solà C, Burgos M, Plazuelo A, Toja J, Plans M, Prat N. Heavy metalbioaccumulation and macroinvertebrate community changes in aMediterranean stream affected by acid mine drainage and anaccidental spill (Guadiamar River, SW Spain). Sci Total Environ2004;333:109–26.

Staessen J, Yeoman N, Fletcher A, Markowe HL, Marmot MG, RoseG, et al. Blood cadmium in London civil servants. Int J Epidemiol1990;19:362.

57F. Gil et al. / Science of the Total Environment 372 (2006) 49–57

Taggart MA, Carlisle M, Pain DJ, Williams R, Green D, Osborn D, etal. Arsenic levels in the soils and macrophytes of the “Entremuros”after the Aznalcóllar mine spill. Environ Pollut 2005;133:129–38.

Taggart MA, Figuerola J, Green AJ, Mateo R, Deacon C, Osborn D, etal. After the Aznalcóllar mine spill: arsenic, zinc, selenium, leadand copper levels in the livers and bones of five waterfowl species.Environ Res 2006;100:349–61.

Vidal M, López-Sánchez JF, Sastre J, Jiménez G, Dagnac T, Rubio R,et al. Prediction of the impact of the Aznalcóllar toxic spill on thetrace element contamination of agricultural soils. Sci Total Environ1999;242:131–48.