Environmental Research 141 (2015) 58–68

Contents lists available at ScienceDirect

Environmental Research
journal homepage: www.elsevier.com/locate/envres

Fish consumption patterns and hair mercury levels in children and

their mothers in 17 EU countries
Argelia Castaño a,n, Francisco Cutanda a, Marta Esteban a, Peter Pärt b, Carmen Navarro a,
Silvia Gómez a, Montserrat Rosado a, Ana López a, Estrella López c, Karen Exley d,
Birgit K. Schindler e, Eva Govarts f, Ludwine Casteleyn g, Marike Kolossa-Gehring h,
Ulrike Fiddicke h, Holger Koch e, Jürgen Angerer e, Elly Den Hond f, Greet Schoeters f,
Ovnair Sepai d, Milena Horvat i, Lisbeth E. Knudsen j, Dominique Aerts k, Anke Joas l,
Pierre Biot k, Reinhard Joas l, José A. Jiménez-Guerrero a, Gema Diaz a, Catherine Pirard m,
Andromachi Katsonouri n, Milena Cerna o, Arno C. Gutleb p, Danuta Ligocka q,
Fátima M Reis r, Marika Berglund s, Ioana-Rodica Lupsa t, Katarína Halzlová u,
Corinne Charlier m, Elizabeth Cullen v, Adamos Hadjipanayis w, Andrea Krsková o,
Janne F Jensen j, Jeanette K Nielsen j, Gerda Schwedler h, Michael Wilhelm x, Peter Rudnai y,
Szilvia Középesy y, Fred Davidson z, Mark E. Fischer aa, Beata Janasik q, Sónia Namorado r,
Anca E. Gurzau t, Michal Jajcaj u, Darja Mazej i, Janja Snoj Tratnik i, Kristin Larsson s,
Andrea Lehmann ab, Pierre Crettaz ab, Giagkos Lavranos n, Manuel Posada c
a
Environmental Toxicology, Centro Nacional de Sanidad Ambiental (CNSA), Instituto de Salud Carlos III(ISCIII), 28220 Majadahonda, Madrid, Spain
b
European Commission, Joint Research Centre (JRC), Institute of Environment and Sustainability, 21027 Ispra, Italy
c
Instituto de Investigacion de Enfermedades Raras (IIER) Instituto de Salud Carlos III (ISCIII), Madrid, Spain
d
Public Health England, Centre for Radiation, Chemical and Environmental Hazards, Chilton, United Kingdom
e
Institute for Prevention and Occupational Medicine of the German Social Accident Insurance-Institute of the Ruhr-Universität Bochum (IPA), Germany
f
Environmental Risk and Health, Flemish Institute for Technological Research (VITO), Mol, Belgium
g
University of Leuven, Leuven, Belgium
h
Federal Environment Agency (UBA), Dessau-Rosslau, Berlin, Germany
i
Department of Environmental Sciences, Jožef Stefan Institute, Ljubljana, Slovenia
j
Departament of Public Health, University of Copenhagen, Copenhagen, Demark
k
DG Environment, Federal Public Service Health, Food Chain Safety and Environment, Brussels, Belgium
l
BiPRO, Munich, Germany
m
CHU of Liege, Laboratory of Clinical, Forensic and Environmental Toxicology, Liege, Belgium
n
State General Laboratory, Nicosya, Cyprus
o
National Institute of Public Health, Prague, Czech Republic
p
Luxembourg Institute of Science and Technology (LIST) Louxembourg
q
Nofer Institute of Occupational Medicine, Lodz, Poland
r
Lisbon Faculty of Medicine, Lisbon, Portugal
s
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
t
Environmental Health Center, Cluj-Napoca, Romania
u
Public Health Authority of the Slovak Republic (UVZ SR), Bratislava, Slovak Republic
v
Department of Community of Health, Health Service Executive, Kildare, Ireland
w
Larnaca General Hospital, Larnaca, Cyprus
x
Department of Hygiene, Social and Environmental Medicine, Ruhr-University Bochum, Germany
y
National Institute of Environmental Health, Budapest, Hungary
z
Public Analyst's Laboratory Health Service Executive, Cork, Ireland
aa
Laboratoire National de Santé, Dudelange, Luxembourg
ab
Federal Office of Public Health (FOPH), Berne, Switzerland

Abbreviations: COPHES, COnsortium to Perform Human biomonitoring on a European Scale; CVD, Cardiovascular disease; DEMOCOPHES, DEMOnstration of a study to
COordinate and Perform Human biomonitoring on a European Scale; HBM, Human Biomonitoring; MeHg, methylmercury; PUFAs, polyunsaturated fatty acids

http://dx.doi.org/10.1016/j.envres.2014.10.029
0013-9351/& 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
 A. Castaño et al. / Environmental Research 141 (2015) 58–68 59

art ic l e i nf o a b s t r a c t

Article history: The toxicity of methylmercury (MeHg) in humans is well established and the main source of exposure is
Received 21 May 2014 via the consumption of large marine fish and mammals. Of particular concern are the potential neuro-
Received in revised form developmental effects of early life exposure to low-levels of MeHg. Therefore, it is important that pregnant
17 October 2014
women, children and women of childbearing age are, as far as possible, protected from MeHg exposure.
Accepted 20 October 2014
Within the European project DEMOCOPHES, we have analyzed mercury (Hg) in hair in 1799 mother–
Available online 7 February 2015
child pairs from 17 European countries using a strictly harmonized protocol for mercury analysis. Parallel,
Keywords: harmonized questionnaires on dietary habits provided information on consumption patterns of fish and
Human Biomonitoring marine products. After hierarchical cluster analysis of consumption habits of the mother–child pairs, the
Mercury in hair DEMOCOPHES cohort can be classified into two branches of approximately similar size: one with high fish
Sea fish
consumption (H) and another with low consumption (L). All countries have representatives in both
Shellfish
branches, but Belgium, Denmark, Spain, Portugal and Sweden have twice as many or more mother–child
Seafood products
pairs in H than in L. For Switzerland, Czech Republic, Hungary, Poland, Romania, Slovenia and Slovakia the
situation is the opposite, with more representatives in L than H.
There is a strong correlation (r¼ 0.72) in hair mercury concentration between the mother and child in
the same family, which indicates that they have a similar exposure situation. The clustering of mother–
child pairs on basis of their fish consumption revealed some interesting patterns. One is that for the same
sea fish consumption, other food items of marine origin, like seafood products or shellfish, contribute
significantly to the mercury levels in hair. We conclude that additional studies are needed to assess and
quantify exposure to mercury from seafood products, in particular. The cluster analysis also showed that
95% of mothers who consume once per week fish only, and no other marine products, have mercury levels
0.55 μg/g. Thus, the 95th percentile of the distribution in this group is only around half the US-EPA re-
commended threshold of 1 μg/g mercury in hair. Consumption of freshwater fish played a minor role in
contributing to mercury exposure in the studied cohort.
The DEMOCOPHES data shows that there are significant differences in MeHg exposure across the EU
and that exposure is highly correlated with consumption of fish and marine products. Fish and marine
products are key components of a healthy human diet and are important both traditionally and culturally
in many parts of Europe. Therefore, the communication of the potential risks of mercury exposure needs to
be carefully balanced to take into account traditional and cultural values as well as the potential health
benefits from fish consumption. European harmonized human biomonitoring programs provide an addi-
tional dimension to national HMB programs and can assist national authorities to tailor mitigation and
adaptation strategies (dietary advice, risk communication, etc.) to their country’s specific requirements.
& 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction The neurotoxicity of MeHg in humans is well established. The

vulnerability of the developing fetus to MeHg exposure resulted in
Lifestyles and particularly the diet play a crucial role in personal extreme fetal neurotoxicity as first described for Minamata, Japan,
exposure to environmental chemicals. Exposure to methylmercury in 1956 (Kurland et al., 1960; Yorifuji et al., 2013). In Minamata, the
(MeHg) is a clear example as the general population is exposed to exposure levels were high. Later, several large-scale epidemiolo-
this metal compound through consumption of fish and other pro- gical studies (New Zealand, the Seychelles and Faroe Islands) have
ducts from the aquatic environment (Schoeman et al., 2009). shown that even low-level exposure to MeHg during early life
Mercury is a ubiquitous heavy metal, naturally present in the stages could interfere in neural development leading to cognitive
environment but human activities have increased its concentra- malfunction (reduction in IQ, attention deficit disorder) later in life
tion in the environment about three-fold over the last century (Kjellstrom et al., 1986; Julshamn et al., 1987; Myers et al., 2003).
(Mason et al., 2012; Lamborg et al., 2014). In aquatic ecosystems, On the other hand, negative effects of contaminants could be
mercury is transformed to its organic form, MeHg, which is more counter-balanced by the positive effects of healthy nutrients in fish
and marine products (Karagas et al., 2012; Choi et al., 2014). This
bioavailable and bioaccumulates in aquatic food chains to reach
highlights the importance of balancing the risks and benefits of
the highest concentrations in the upper trophic levels. In October
fish consumption.
2013, the Minamata Convention under the auspices of the United
Since the symptoms of methylmercury exposure are subtle and
Nations Environment Program (UNEP), a global action to protect
multi-causal, there is still no consensus on a heath-based guidance
human health and the environment from anthropogenic emissions value for MeHg exposure despite the large number of recent studies
and releases of mercury and mercury compounds, was signed trying to connect low exposure levels to actual risk (Schoeman
(http://www.mercuryconvention.org). Although mercury emis- et al., 2009; Karagas et al., 2012; Valent et al., 2013). However, there
sions will be reduced when this treaty becomes fully operative, old is a general recommendation that pregnant women, children and
mercury releases in deposits (e.g. in soil, sediments, ice) will still women of childbearing age should be protected as much as possible
be mobilized and become biologically available. Therefore, old from mercury exposure. Therefore, it is important to know what the
“legacy” contamination will contribute to the current mercury actual exposure is to MeHg in the general population and what the
levels in the environment and consequently determine human sources of exposure are in order to formulate adequate mitigation
exposure for many years to come (Selin, 2009). strategies and recommendations.

n
Corresponding author.
E-mail address: castano@isciii.es (A. Castaño).
60 A. Castaño et al. / Environmental Research 141 (2015) 58–68

young families, and managing the delicate balance between the

assumed health benefits of a diet rich in fish and the potential
negative effects of an increased burden of environmental con-
taminants including mercury.

2. Materials and methods

DEMOCOPHES was a pilot study involving 17 European coun-

tries (Fig. 1). The aim of the project was to test the feasibility of a
harmonized human biomonitoring approach; testing protocols
developed in the EU project COPHES (Joas et al., 2012).

2.1. Harmonized tools

2.1.1. The European protocol

Fig. 1. Map with the urban (●) and rural (X) sampling locations in the 17 European
Based on the European study protocol, countries developed their
implementing countries of DEMOCOPHES.
national protocols with only minor adaptations so as not to jeo-
pardize the comparability of the results. The study design has been
described previously by Becker et al. (2014). In summary, 120 mo-
The European project DEMOCOPHES (DEMOnstration of a study
ther–child pairs, up to 45 years and from 6 to 11 years, respectively,
to COordinate and Perform Human biomonitoring on a European
in each country (except Cyprus and Luxembourg which were ex-
Scale) was initiated to demonstrate how harmonized HBM tools
pected to contribute with 60 pairs per country because of their
developed in the European project COPHES (Consortium to Per-
smaller population size) were to be recruited from two locations
form Human biomonitoring in a European Scale) could be applied
(Fig. 1) representing the upper and lower degree of urbanization
in Europe to provide baseline information on selected con-
(urban and rural), via inhabitant registries or schools. Participants
taminant levels in the European population. Mercury was one of
were recruited between September 2011 and February 2012 and
the contaminants included and here we give an insight into
had been living for the 5 last years or more in the same location.
mercury levels, based on hair measurements, of children and their
mothers in 17 European countries measured under strictly stan- Volunteers living in hospitals or institutions, who were homeless, or
dardized and harmonized conditions (Esteban et al., 2015). Our presenting metabolic disturbances (e.g. diabetes, nephritic syn-
results build on analysis of total mercury in hair, which is sig- drome or porphyria) or abnormal urine excretions (creatinine va-
nificantly cheaper than analysis of methylmercury. For hair ana- lues o0.3 g/L or 43 g/L) were excluded (WHO, 1996).
lysis, total mercury is generally accepted as a good proxy for MeHg
exposure (Harkins and Susten, 2003). 2.1.2. Samples and data collection
As shown before, DEMOCOPHES results confirmed the clear Hair samples were collected according to the instructions de-
correlation between consumption of fish and marine products and tailed in the Standard Operation Procedure (SOP) included in the
mercury levels in hair (Den Hond et al., 2015). However, since the study protocol (Esteban et al., 2015). A trained field worker con-
habits and frequency of fish and marine products consumption ducted the interviews of the mothers based on validated struc-
show great variation among the participating European countries, tured questionnaires. Using different reverse/independent trans-
and since the study applied harmonized questionnaires and ana- lators the quality of translations of the questionnaire into national
lytical procedures, we used this unique opportunity for a more languages was controlled. The questions were categorized into
detailed analysis of the relationship between dietary patterns and following sections: residence environment, nutrition, smoking
mercury exposure. In addition, the fact that the study included behavior, occupation and socio-demography. The mothers an-
mother–child pairs enabled us to analyze how adult food habits swered all questions including those concerning the consumption
and related exposure influences children’s exposures in the same of fish and marine products for both herself and her child (Fig. 2).
family. This type of information is pivotal for health and food Each question was asked separately for the mother and for her
safety authorities when developing dietary recommendations for child, so that personalized data could be collected.

Fig. 2. Image of the harmonized DEMOCOPHES questionnaire on consumption habits of fish and marine products.
 A. Castaño et al. / Environmental Research 141 (2015) 58–68 61

Additionally, data on the sampling procedure and basic in- Table 1

formation about the samples were collected e.g. whether the hair Number of mother–child pairs recruited in each DEMOCOPHES implementing
country and number of pairs with complete fish questionnaires (valid for analysis).
had been dyed or tinted within the previous 6 months.
Country Number of pairs Number of valid pairs
2.1.3. Chemical analysis
Mercury determinations in the DEMOCOPHES pilot study were Belgium 129 126
Cyprus 60 59
done in laboratories which successfully passed the COPHES/DE- Czech Republic 120 120
MOCOPHES external quality assurance program (Esteban et al., Denmark 145 138
2015 and Table S1 as Supplementary material). To assess mercury Germany 120 120
Hungary 120 120
in hair, the closest centimeters (maximum 3 cm) to the scalp of the
Ireland 120 101
hair samples were analyzed. Luxembourg 60 52
Poland 120 116
2.1.4. Databases management Portugal 120 120
The analytical results and the information gathered with the Romania 120 120
Slovakia 129 126
questionnaire were recorded in the national databases using a Slovenia 120 118
common format and coding. Identical quality controls were ap- Spain 134 133
plied at national level, using a centrally developed script (written Sweden 100 94
Switzerland 120 115
in R statistical software, R Development CoreTeam, 2012). The data
United Kingdom 21 21
from all the countries were merged into the European database
(Den Hond et al., 2015).
2.3. Analysis of mercury levels in hair in relation to diet patterns
2.1.5. Training
In support of the study protocol and the Standard Operating Once mother–child pairs were classified, and consumption
Procedures, various training sessions were organized covering the patterns were described, mercury concentrations in hair for the
following tasks: sampling, sample transport and preservation, in- different groups and diets were compared. Hair mercury levels of
terview/questionnaire conduct, statistical analysis, ethics and mothers and children were taken from the central database (Den
communication (Fiddicke et al., 2015). Hond et al., 2015). Mercury levels have been validated and stored
in the central database after imputation of values below the limits
2.1.6. Ethical approval of quantification (LOQ) as LOQ/2. Medians and 90th percentiles of
The ethical committees of each involved European country
mercury in hair were calculated for the different groups. Wilcox-
approved the study protocol and country specific requirements
on–Mann–Whitney rank sum tests were used in order to compare
were followed. All mothers gave written informed consent on her consumption of food types between groups, and Kendall's tau-b
and her child's behalf (Casteleyn et al., 2015) (www.eu-hbm.info/ correlation was utilized to analyze the association of each type
democophes). between mothers and children. It was also used to measure the
association of mercury levels between mother and child.
2.2. Statistical analysis of data The clustering of the European mother–child pairs on the
dietary answers was the basis for the analysis of mercury exposure
2.2.1. Cluster analysis of diet-patterns in the European sample (from hair analysis) in relation to consumption of food items from
The classification procedure (cluster analysis) was based on the the marine or freshwater environments. All pairwise comparisons
seven possible distinct answers to questions B, C, D, and E (SEA FISH; between cluster groups were tested for differences in diet and
SHELLFISH; FRESHWATER FISH and SEAFOOD PRODUCTS) (Fig. 2) from the mother– mercury (Wilcoxon–Mann–Whitney test). When comparing
child pair, totaling eight variables with seven levels. groups, two kinds of results were of interest for the discussion of
DEMOCOPHES data including answers to the questionnaire on mercury exposure factors in this work:
fish consumption, and number of amalgam fillings of mothers and
children along with area of residence, gender, age, and parental 1. Equal SEA FISH consumption/significantly different Hg hair levels.
educational levels were taken from the central database (Den 2. Significantly different SEA FISH consumptions/equal Hg hair level.
Hond et al., 2015).
Hierarchical cluster analysis, and the complete linkage algo- The confidence level for hypothesis testing was 0.05. All sta-
rithm of cluster building, was used, with a Manhattan distance as tistical analyzes were carried out using Stata 12 (Stata Corp LP,
dissimilarity measure (Johnson and Wichern, 1988). Manhattan USA). Automated searches, written as Stata do scripts, were used
dissimilarity would be zero if two questionnaires answers were in order to find diet-patterns of high frequency in the European
identical (all eight variables). It would be one if just one of the sample, and within countries.
eight variables differs in one step in the sequence “several times
per day” - “daily” - “several times per week” - “once per week” -
“several times per month” - “once per month” - “almost never”. It 3. Results
would be two if that happens in just two items in the ques-
tionnaire, or if there is a two steps difference in just one item, and Most of the 17 DEMOCOPHES implementing countries met the
so on. In general, the dissimilarity will be the total number of steps minimum required sample size of 120 mother child pairs (60 in
of difference between two questionnaires there are for all items Luxembourg and Cyprus because of their smaller population size).
(Fig. S1 in Supplementary material, shows an example for two Exceptions were UK (21 pairs) and to a minor extent Sweden (100
items). pairs). Some countries extended their sample sizes beyond the
The analysis of combined consumption in mother–child pairs minimal DEMOCOPHES requirements (Table 1). The grand total
was based on the lowest reported consumption in the pair. was 1858 pairs.
62 A. Castaño et al. / Environmental Research 141 (2015) 58–68

Fig. 3. Dendrogram of the hierarchical cluster analysis for fish and marine product consumption of participants in the European sample DEMOCOPHES with the two main
branches H (high consumption) and L (low consumption) and the secondary groups at the third level of dendrogram: H1, H2, H3 and L1, L2. Bar graph below each group,
showing percentages of pairs consuming SEA FISH more than weekly, SHELLFISH more than monthly, SEAFOOD PRODUCTS more than monthly and FRESHWATER FISH more than monthly.

A total of 59 pairs were excluded from the classification of fish total) and “low fish consumption” (L) branch with 990 pairs (55%
products consumption because of their incomplete questionnaires. of the total). All the top scoring pairs for any of the four questions
In Ireland, 19 pairs were excluded, because SEA FISH was consistently are in H. Moreover, considering a family habit of once per week as
reported, but there were missing answers for the rest of the items. a reference for SEA FISH (both members of the pair) and once per
The rest of the exclusions are scattered among eleven countries. month as a reference for SHELLFISH, FRESHWATER FISH and SEAFOOD, it was
Thus a final set of 1799 pairs was considered in the present ana- found that pairs reporting higher frequencies than the reference
lysis (Table 1). were all in H for SEA FISH (except three), they were ten times more
Mercury levels in the hair of participants showed large differ- abundant in H than in L for SHELLFISH and they were twice as many
ences, up to a factor of 40 in national medians, among the 17 in H than in L for SEAFOOD. Although FRESHWATER FISH presented just
participating countries. Analysis of the data clearly indicates that 1.4 more pairs above the reference in H, their consumption levels
differences in mercury levels for the European participants are were higher, even above one per week. Pairs with frequencies
associated with diet (Den Hond et al., 2015 and Tables S2 and S3 as above the reference for more than one question were mostly in H.
Supplementary material). All countries had participants in both branches, but it can be
pointed out that Belgium, Denmark, Spain, Portugal and Sweden
3.1. Cluster analysis of diet had twice as many or more pairs in H (high fish consumption)
than in L (the low fish consumption) groups, and vice versa for
Fish consumption reported in the DEMOCOPHES ques- Switzerland, Czech Republic, Hungary, Poland, Romania, Slovenia
tionnaires can be classified in two main branches and in secondary and Slovakia (Table S4).
groups. Fig. 3 and Fig. S2 show the dendrogram of the hierarchical There are no significant differences between the assignment of
cluster analysis considering the eight variables (questions B, C, D, pairs to H or L between the rural and urban locations of each
and E for each member of the pair). First, as a result of the clas- country (p 40.110), with the exceptions of Hungary (p ¼0.024) and
sification procedure, two broad branches were identified: “high Slovenia (p ¼0.005) where rural locations show lower consump-
fish consumption” (H) branch containing 809 pairs (45% of the tion levels.

Table 2
Recurrent diet patterns in pairs (both mother and child) of groups H3 and L2, according to the answers given to the DEMOCOPHES fish and marine product consumption
questionnaire. SHELLFISH and SEAFOOD PRODUCTS answers for both members of the pair are “almost never” in all six diet patterns. Mercury in hair median, 90th percentile and
percentage of subjects above the limit of quantification (LOQ) are also shown.

I II III IV V VI

Groupa L2 L2 L2 L2 H3 H3
N (pairs) 128 37 80 113 106 44
SEA FISH Mother/Child (Almost) never (Almost) never Once a month 2–3 Times per month Once a week Several times/week
FRESH WATER FISH Mother/Child (Almost) never Once a month (Almost) never (Almost) never (Almost) never (Almost) never
Hg in hair (lg/g) Mother P50 0.069 0.069 0.118 0.127 0.149 0.808
P90 0.243 0.345 0.387 0.479 0.550 2.391
4 LOQ % 53.5 49.5 59.8 74.3 75.0 84.9
Hg in hair (lg/g) Child P50 0.061 oLOQ 0.078 0.087 0.094 0.766
P90 0.168 0.178 0.285 0.472 0.473 1.918
4 LOQ % 52.3 49.4 38.0 68.0 73.5 74.5

N ¼number of pairs. P50 ¼50th percentile. P90¼ 90th percentile.

a
Group according to cluster classification shown in Fig. 2.
 A. Castaño et al. / Environmental Research 141 (2015) 58–68 63

Table 3 Likewise, the low fish consumption branch L was divided into a
Correlation (Kendall's tau-b) between mothers' and children's frequency of con- small group (L1, 72 pairs) and a mainstream group (L2, 918 pairs).
sumption of SEA FISH, SHELLFISH, SEAFOOD PRODUCTS and FRESHWATER FISH and mercury in hair
Within the L branch, the group L1 includes pairs with an in-
(μg/g). Correlation within each of the main groups according to cluster classification
represented in Fig. 2 (H1,H2, H3, L1 and L2). termediate range of SEA FISH consumption. None of the pairs an-
swered “almost never” to all B, C, D and E questions in L1. The
Group N SEA FISH SHELLFISH SEAFOOD PRODUCTS FRESHWATER FISH Hg in hair most prominent feature within the L groups is the difference in
n n n n consumption of SEAFOOD PRODUCTS: in L1, 21 pairs out of 72 (29%) had
H1 102 0.599 0.465 0.621 0.649 0.655n
H2 44 0.693n 0.445n 0.531n  0.048 0.592n consumption levels above the reference of once per month,
H3 663 0.667n 0.536n 0.509n 0.716n 0.535n whereas there were none in L2.
L1 72 0.283n 0.185n  0.010 0.427n 0.442n L2 grouped participants with the lowest consumption in the
L2 918 0.424n 0.589n 0.455n 0.699n 0.429n
whole European sample. Half the pairs in the group answered “2
Total 1858 0.678n 0.598n 0.610n 0.748n 0.545n
or 3 times per month” or less for SEA FISH AND “almost never” for
n
po 0.05. consumption of SHELLFISH, SEAFOOD PRODUCTS and FRESHWATER FISH.

3.2. Group comparison and analysis.

The two big branches (H and L) were further divided in a main
classification of three (H1; H2; H3) and two groups (L1; L2) of
A distinct pattern could be observed in the analysis of ques-
unequal size but optimal dissimilarity (Fig. 3).
tionnaire replies. In the European sample, six recurring ques-
The high consumption branch (H) includes a big mainstream
tionnaire answers for mother and child pairs (Table 2) were
group (H3, 663 pairs) in addition to two smaller extreme groups
dominating. Given that the theoretical number of answers is
(H1 and H2) with high consumption (Fig. 3). H1 (102 pairs) and H2 5.7 million, six recurrent diet patterns show that there is a clear
(44 pairs) stand out from H3 because of their higher consumption preference in consumption patterns of fish and marine products in
rates of SEA FISH and, additionally, FRESHWATER FISH (H1) or SHELLFISH and the studied European cohort. The six recurring answers are in core
SEAFOOD PRODUCTS (H2). Therefore, H1 and H2 gathered pairs with the groups H3 and L2 and account for little less than a third of the
highest scores in SEA FISH, SHELLFISH, SEAFOOD PRODUCTS and FRESHWATER FISH total number of replies. Thus, 358 pairs in L2 (39%) and 150 pairs
consumption. in H3 (23%) replied just one of the six recurring fish patterns

Table 4
Results for mercury in hair (μg/g) for mothers and children in each of the groups, and selected subgroups utilized for comparisons. Groups were obtained from cluster
analysis of fish and marine products diet. Median (P50), 90th percentile (P90) and percentage of samples above the limit of quantification (LOQ).

Mothers Children

Group (N) Subgroup (N) P50 P90 Max 4LOQ % P50 95% CI P90 95%CI Max 4LOQ %
95% CI 95% CI
H1 (102) 0.630 2.770 8.90 97.0 0.370 1.560 6.60 93.1
0.442–0.877 2.000–3.400 0.280–0.494 1.164–2.764
H1a (88) 0.501 2.100 8.90 96.6 0.310 1.300 6.60 92.0
0.331–0.770 1.515–3.592 0.209–0.406 0.875–2.390
H1a1 (71) 0.400 1.671 4.50 95.7 0.232 0.898 2.80 90.0
0.233–0.683 1.305–2.094 0.168–0.345 0.702–1.791
H1b (14) 1.724 3.400 3.40 100 0.950 3.700 4.20 100
1.211–3.217 2.871–3.400 0.645–1.433 1.400–4.200

H2 (44) 1.335 3.200 6.70 100 0.979 2.103 7.10 100

1.003–1.797 2.600–5.732 0.600–1.156 1.593–5.841

H3 (663) 0.404 1.649 9.66 97.9 0.263 1.200 6.40 94.5

0.352–0.453 1.504–1.892 0.229–0.289 1.078–1.335
H3a (506) 0.347 1.619 8.90 97.3 0.220 1.188 6.40 93.6
0.311–0.395 1.444–1.800 0.205–0.259 1.016–1.300
H3a1 (291) 0.347 1.818 8.90 96.9 0.220 1.266 6.40 98.8
0.282–0.413 1.539–2.294 0.198–0.284 1.171–1.681
H3b (107) 0.587 2.434 9.66 100 0.356 1.553 5.07 97.2
0.455–0.790 1.606–3.362 0.304–0.469 1.102–2.009
H3c (50) 0.605 1.449 2.70 100 0.297 0.967 2.90 98.0
0.440–0.737 0.987–2.154 0.228–0.348 0.531–2.538

L1 (72) 0.333 1.470 3.90 95.8 0.151 0.700 1.30 88.7

0.241–0.451 0.997–2.731 0.116–0.199 0.381–1.063
L1a (63) 0.340 1.320 3.90 95.2 0.137 0.700 1.30 87.1
0.226–0.451 0.915–2.472 0.095–0.176 0.385–1.090
L1b (9) 0.250 2.822 2.82 100 0.205 0.490 0.49 100
0.186–1.432 0.610–2.822 0.156–0.279 0.263–0.490

L2 (918) 0.132 0.520 3.90 83.7 0.076 0.355 3.20 78.3

0.120–0.148 0.457–0.601 0.069–0.088 0.310–0.404
L2a (848) 0.130 0.518 3.90 84.4 0.074 0.352 3.20 79.0
0.118–0.146 0.449–0.597 0.069–0.087 0.310–0.414
L2b (64) 0.143 0.518 1.21 73.0 0.080 0.341 1.82 66.7
0.081–0.196 0.380–0.871 0.069–0.140 0.254–0.972
L2c (6) 0.339 1.000 1.00 100 0.208 0.900 0.90 100
0.161–0.969 0.457–1.000 0.130–0.848 0.274–0.900

N ¼number of pairs, P50¼ 50th percentile, P90¼ 90th percentile, and CI ¼ confidence interval.
64 A. Castaño et al. / Environmental Research 141 (2015) 58–68

shown in Table 2. The common denominator in these groups is the Table 5

reply “almost never” to the consumption of SHELLFISH and SEAFOOD Comparison of subgroups H2 and H3b (A); H3a1 and H3b (B) and H1a1 and H3a1
(C) with results of the tests of equality of distribution of SEA FISH, SHELLFISH, SEAFOOD
PRODUCTS. This makes it possible to compare mercury levels just on
PRODUCTS, FRESHWATER FISH and mercury in hair between both subgroups (Wilcoxon–
the basis of fish (seawater or fresh water) consumption without Mann–Whitney rank sum test). Percentages of mothers or children having amal-
interference of SHELLFISH or SEAFOOD PRODUCTS. gam fillings, consuming SEA FISH and SHELLFISH more than weekly, or SEAFOOD PRODUCTS and
FRESHWATER FISH more than monthly, along with median values of mercury in hair (mg/
None of the recurring fish patterns identified in the ques-
g) are shown to illustrate the differences between subgroups.
tionnaires seems to be characteristic for a particular country. There
are some countries with up to four prevalent reply patterns, while A
in other countries they are rare or even absent, but considering the
lack of country representativeness of the samples in general, the
Subgroups
information should be considered cautiously. For example, pattern
I (Table 2) is prominent in Romania and Hungary but absent in H3b H2 p
(N¼ 107) (N¼ 44)
Cyprus, Spain, Luxembourg, Portugal and United Kingdom. Look-
ing at Romanian and Hungarian pairs, there is a strong incidence Amalgam fillings Mothers 67.7% 66.7% 0.910
of patterns I, II and III with pattern II (FRESHWATER FISH) being char- Children 4.7% 11.4% 0.139
acteristic for these two countries. Patterns I, III, and IV of low SEA FISH Mothers 44.9% 61.3% 0.201
( 41/week) Children 43.0% 59.1% 0.137
consumption are frequent in Switzerland, Czech Republic, Poland
SHELLFISH Mothers 89.7% 68.2% o 0.001n
and Slovenia. In Germany and Slovak Republic patterns I, II, IV and ( 41/month) Children 43.0% 45.5% 0.674
V make up half of the total DEMOCOPHES sample. Pattern VI ap- SEAFOOD PRODUCTS Mothers 3.7% 100% o 0.001n
plies only to Portugal. Pairs in Belgium, Denmark, Spain and ( 41/month) Children 0.0% 88.6% o 0.001n
FRESH WATER FISH Mothers 0.0% 6.8% 0.806
Sweden rarely provide any recurrent combination of answers to ( 41/month) Children 0.0% 0.0% 0.099
the questionnaire and even less frequently agree in any combi- Mercury Mothers 0.587 1.335 o 0.001n
nation (i.e. four repetitions at most in each country) (Table S4). (median lg/g) Children 0.356 0.979 o 0.001n
B
The five main groups (H1, H2, H3, L1 and L2) differ significantly
(p o0.05) in each of the four questions, when comparing every Subgroups
two groups, except in the low consumption branches (L1 and L2)
that have the same consumption of SHELLFISH (p ¼0.862) and FRESH- H3a1 H3b p
(N¼ 291) (N¼ 107)
WATER FISH (p ¼0.725).
Comparing the questionnaire replies of mothers and their
Amalgam fillings Mothers 69.7% 67.7% 0.702
children, about half of the pairs in DEMOCOPHES report identical Children 9.6% 4.7% 0.118
consumption patterns (880 pairs). Still, the mother–child replies SEA FISH Mothers 46.7% 44.9% 0.737
are positively correlated for all except two items FRESHWATER FISH in ( 41/week) Children 45.7% 43.0% 0.383
SHELLFISH Mothers 4.1% 89.7% o 0.001n
the H2 group and SEAFOOD PRODUCTS in L1 (Table 3). In fact L1-the
( 41/month) Children 0.7% 43.0% o 0.001n
highest consumer group in the lowest branch-has the lowest po- SEAFOOD PRODUCTS Mothers 3.1% 3.7% 0.012n
sitive correlation (Kendall's tau-b) between mothers' and chil- ( 41/month) Children 0.3% 0.0% 0.325
dren's replies. The largest differences between mother and child FRESH WATER FISH Mothers 0.3% 0.0% 0.357
( 41/month) Children 2.4% 0.0% 0.168
answers to the same item occurred in this group (L1). In general, Mercury Mothers 0.347 0.587 o 0.001n
mothers eat fish and marine products more frequently than their (median lg/g) Children 0.220 0.356 0.003n
children, particularly SEAFOOD PRODUCTS with 15.3% of mothers in L1 C

exceeding their children's frequency of consumption by more than

three levels. Subgroups
When comparing questionnaires replies for mothers and
questionnaires replies for children independently, the five main H3a1 H1a1 p
(N¼ 291) (N¼ 71)
groups (H1, H2, H3, L1 and L2) also differ significantly (po 0.05) in
each of the four questions when comparing every two groups, as
Amalgam fillings Mothers 69.7% 73.2% 0.558
was shown for the pairs. But in this case the exceptions increases Children 9.6% 5.6% 0.289
to five cases for mothers (SEA FISH for H1–H2; SHELLFISH for H1–L1; SEA FISH Mothers 46.7% 26.8% 0.018n
SEAFOOD PRODUCTS for H2–L1 and FRESHWATER FISH for H2–H3 and L1–L2)
( 41/week) Children 45.7% 25.4% 0.001n
SHELLFISH Mothers 4.1% 12.7% 0.508
and in two more for children (SHELLFISH and FRESHWATER FISH for L1–L2). ( 41/month) Children 0.7% 2.8% 0.476
In other words, mothers in the two extreme high consumption H SEAFOOD PRODUCTS Mothers 3.1% 12.7% 0.019n
branches reported the same frequency of SEA FISH consumption ( 41/month) Children 0.3% 0.0% 0.173
FRESH WATER FISH Mothers 0.3% 98.6% o 0.001n
(p ¼0.072); and, children in the low consumption branches (L1
( 41/month) Children 2.4% 100.0% o 0.001n
and L2) replied the same for SHELLFISH (p ¼0.898), 93% of them re- Mercury Mothers 0.347 0.400 0.598
ported “almost never” to question C. (median lg/g) Children 0.220 0.232 0.664
When levels of mercury in hair were compared to the classi-
N¼ number of pairs.
fication based on fish consumption (Fig. 3), the dendrogram n
p o 0.05.
matched differences in mercury levels both in mothers and chil-
dren. The main five groups (level 2 on Fig. 3) differed in their
mercury values in hair, which are shown in Table 4. H1, with a (1.3 mg/g for mothers and 0.98 mg/g for children) and the highest
high consumption of SEA FISH and FRESH FRESHWATER FISH, showed ex- 90th percentiles (3.2 mg/g for mothers and 2.1 mg/g for children).
treme values of up to 8.9 mg/g mercury in one mother and 6.6 mg/g H groups have higher medians (P50) and 90th percentiles (P90)
mercury in one child. Group H2, with a high consumption of than L ones, although the highest exposure pairs in L1 clearly keep
SHELLFISH and SEAFOOD PRODUCTS, shows the highest median levels up with typical mercury levels in H3.
 A. Castaño et al. / Environmental Research 141 (2015) 58–68 65

H3 group, representing the 37% of the studied population, consumption (almost everyone in H1b and almost nobody in H1a
consumed fish and/or marine products (B, C, D, E questions) on a having SEAFOOD PRODUCTS monthly) while the consumption frequency
weekly basis and its median mercury values are 0.4 μg/g. In fact of SEA FISH is the same, and 3 times more mercury in hair both in
except for the H2 group, the P50 values for the five main groups mothers and their children of H2 with regard to H3a1 or of H1b
were below 1 μg/g mercury in hair. with regard to H1a (Table 4).
Median as well as P90 values are higher for mothers than for Consumption of SHELLFISH also shows a certain influence on the
children. The correlation between mother and child mercury levels increase of Hg levels as was observed in a number of pair-wise
was a bit lower in low consumption groups than in higher con- group comparisons. To exemplify this, two subgroups of H3 (H3a1
sumption ones (Table 3). Correlation between mother and child in and H3b) were compared (Table 5B). When comparing H3a1 and
H3b SHELLFISH consumption is significantly different, and SEA FISH and
mercury shows wider variation by country (range 0.195–0.494)
FRESHWATER FISH are consumed at the same level. Mercury in hair was
than by diet group.
significantly higher in the subgroup with higher SHELLFISH con-
Mercury levels increased in relation to the frequency of SEA FISH
sumption (Table 5B).
consumption in study participants reporting no consumption of
Therefore consumption of SHELLFISH and SEAFOOD PRODUCTS need to
SHELLFISH and SEAFOOD PRODUCTS (Table 2). 90% of the women consuming
be further investigated, in addition to already established con-
only SEA FISH once per week (pattern V) (independently of the tribution of SEA FISH with respect to mercury exposure.
species or the size) had mercury levels of 0.55 μg/g and 50% of the There are also examples in the material of groups that, al-
mothers who consume SEA FISH several times per week (pattern VI) though having significant differences in SEA FISH and FRESHWATER FISH
had mercury levels of 0.80 μg/g. 89% of the mothers in the high consumption, showed no significant differences in mercury levels,
fish consumer group (pattern VI) are at the benchmark value set for example when comparing subgroups H1a1 and H3a1
up by JECFA/WHO 1.9 μg/g (EFSA, 2012) (Table 2). (Table 5C). In this case SEA FISH was consumed almost twice as fre-
All differences in mercury levels between pairs of the five main quently in H3a1 compared to H1a1, with no visible difference in
groups are significant (p o0.05) except for mothers in H3–L1 (the mercury level between the two groups. The consumption of
lowest high consumer group and the highest low consumer group) FRESHWATER FISH (98% more frequent) had no influence on mercury

(p ¼0.153). levels.
H1 and H2 have significantly different levels of mercury in hair The lack of association of FRESHWATER FISH consumption with the
despite having no significant differences in SEA FISH consumption levels of mercury was confirmed in H2 and H1b comparisons. In
and this was the driver to investigate the influence of other dietary this case, SEA FISH and SHELLFISH were equally frequent in the diets of
items on mercury levels. members of both subgroups but there was a significant difference
For that aim, some subgroups of the main five have been se- in the frequency of FRESHWATER FISH in the diet (92% more frequent in
lected from the lower branches of the dendrogram of our cluster H1b both in mothers and children), and no significant effect was
analysis (Table 4). These subgroups show either significantly dif- seen on mercury levels in hair for mothers neither for children
(p ¼0.131 and p ¼0.313, respectively) (Table 4).
ferent mercury levels although the same SEA FISH consumption fre-
quency, or the opposite way, there were significant differences in
SEA FISH consumption but with no differences in mercury in hair. The
4. Discussion
possible contribution of amalgam fillings, their occurrence and
their number, have been tested. Both have been found to be non- In this paper we have compared results both for dietary habits
significant with respect to the mercury levels measured in hair and mercury in hair for the first time in a harmonized way in
(Fisher exact test for the proportion of mothers with amalgam mother–child pairs in 17 European countries. Data are based on
fillings p 40.142, Wilcoxon–Mann–Whitney for their number European dietary habits and on fish marketed and consumed in
p 40.057). Europe. This was possible thanks to the use of a commonly de-
Significant differences in Hg levels although similar SEA FISH, veloped protocol (target population and questionnaires), a full
intake were observed in a number of group-pair comparisons. This training scheme for field work as well as stringent quality control
is exemplified when a subgroup of H3 (H3b) is compared with H2. programs for chemical and data analysis.
Both groups are described in Table 5A, in which the proportion of In the overview of results from the DEMOCOPHES project (Den
mothers and children having each of the food items more fre- Hond et al., 2015), we report that mercury levels in hair of parti-
quently than the reference (once per week for SEA FISH, once per cipants (mother–child pairs) varied with more than a factor of 50
month for SHELLFISH, SEAFOOD PRODUCTS and FRESHWATER FISH), are shown between the lowest (Hungary, geometric mean (GM) 0.02 μg/g
along with median values of mercury in hair. Incidence of amal- hair) and the highest (Portugal, GM 1.03 μg/g hair) with an overall
gam fillings is also shown for reference. The consumption of SEA FISH geometric mean in the DEMOCOPHES material of 0.14 μg/g, based
is high in both groups (50% of the members consume more than on values from the children. The mothers had higher mercury
levels than their children but followed the same pattern. Fish
once per week) with no significant differences between the two
consumption and social status were identified as important and
subgroups (p ¼0.201 for mothers and 0.137 for children). However,
independent determinants of mercury levels, both in mothers and
mercury in hair is more than twice in participants from H2 with
their children, which confirms results from other studies (Ma-
regard to those from H3b. The consumption frequency of SEAFOOD
haffey, 2004). Although France, Greece and Italy did not participate
PRODUCTS is what separates these groups and it seems that SEAFOOD
in the DEMOCOPHES project, recent biomonitoring studies on hair
PRODUCTS is the potential determinant for the increase in the mer-
levels of mercury in women from France (0.60 μg/g), Greece
cury levels. (1.12 μg/g) and Italy (0.77 μg/g) show that they fall well within the
Similar results have been observed in many other pairs of range of the DEMOCOPHES countries in which fish and other
groups, for example between groups H2, and H3a1 with a sig- marine products constitute an important part of the diet (Frery
nificantly (po 0.001) wide gap in their monthly SEAFOOD PRODUCTS et al., 2010; Miklavčič et al., 2013). The wide-spread difference in
consumption (almost everyone in H2 and almost nobody in H3a1 mercury exposure in the European population can be related to
having SEAFOOD PRODUCTS monthly), or between groups H1b and H1a dietary habits and in particular to consumption of fish and other
also with significantly (p o0.001) different SEAFOOD PRODUCTS products from the aquatic environment. With respect to mother–
66 A. Castaño et al. / Environmental Research 141 (2015) 58–68

child pairs, there is a strong correlation (r ¼0.72) in hair mercury product based on several fish species such as Alaska Pollock, Pa-
concentration between the mother and child (Den Hond et al., cific whiting, Tilapia and other species of less market value and it
2015) which shows that there is a common source of exposure in forms the bases for food items such as imitation crab meat, sat-
the studied families. Most likely this is the diet since mercury le- sumi-age/tenpura, hanpen, and fish sausage (Tina et al., 2010).
vels in hair reflect primarily exposure to MeHg from food sources Depending on the origin of surimi products, they may contribute
(Sherman et al., 2013). We also found that mercury from amalgam to the body burden of mercury. There could be a risk that they
fillings is of minor importance (Table 5) with respect to hair levels originate from environments with elevated mercury levels or that
as also reported by others (Sherman et al., 2013). some of the fish species used are top predators in the marine food
An important part of the study was the questionnaire data on chain. Eurostat data from 2011 and 2012 showed an increase of
the participants' consumption frequencies of fish and other pro- 69% in volumes of miscellaneous aquatic products imported from
ducts from the aquatic environment. The weakness of these types extra-EU countries, and it should be mentioned that no imports of
of questionnaires is that they are a retrospective survey in which seaweed and other algae were reported in 2011, but they totalled
the mothers have to remember what and when they ate the dif-
53.000 tonnes in 2012 (EUMOFA, 2014). Clearly, there is a need for
ferent items covered and to what extent their children ate them.
more extensive monitoring of these products to obtain a better
Probably the most certain reply is when the answer is “never” – in
picture of their role as potential sources of mercury exposure.
this case the respondents probably know for sure. Therefore, the
There were also groups of consumers that had lower mercury
questionnaire provided only approximate information on fish
levels than expected on basis of their consumption of SEA FISH and
consumption and was not a precise measure of the fish ingested by
SEAFOOD PRODUCTS. The species of fish consumed were not reported in
an individual. Despite this limitation, the material still allows for
the questionnaires and therefore it is difficult to make exact cor-
analysis of consumption patterns across the DEMOCOPHES coun-
relations. However, it is well known that type of fish, the size, age
tries and of items that may have an additional influence on the
levels of mercury measured in hair. and the position in the food chain are important with respect to
There is a clear pattern in consumption of fish and other mercury content. Top predators accumulate the highest levels
aquatic products across Europe, with relatively higher frequency in since mercury is concentrated along the food chain (NRC, 2000).
the Mediterranean and North European countries and a lower The metal is bioaccumulated over a lifetime and larger and older
frequency in Central European countries. Based on questionnaire specimens will have higher concentrations than younger and
answers we could separate mercury exposure due to SEA FISH con- smaller ones. Therefore active pelagic top predators like tuna,
sumption from FRESHWATER FISH consumption. In the results there was swordfish or long-lived species that such as sharks attain high
no evidence that consumption of FRESHWATER FISH contributes to mercury levels. The mercury values reported can be assumed to
mercury levels in mothers and their children. This finding is not represent exposure from fish landed in Europe by EU fishing fleets.
surprising since the market for FRESHWATER FISH is very limited in The majority of the fish marketed in Europe comes from North-
Europe with 5.3% of total sales of aquatic products while seawater East Atlantic (71.5%) and the Eastern Central Atlantic (13.4%)
fish represents 73.5% and the remaining 21.2% is other types of (EUROSTAT, 2014). According to the European Commission Reg-
aquatic products (EUMOFA, 2014). In some European countries ulation (EC) No. 78/2005 of January 19, 2005 the maximum al-
FRESHWATER FISH could be a significant source of mercury exposure. lowed level of mercury in the species anglerfish, swordfish and
Swedish FRESHWATER FISH (pike, Esox lucius, perch, Perca fluviatilis) tuna is 1 mg/kg. For other fishery products and fish muscle the
contain significant amounts of methylmercury, in the range of 0.5– maximum allowed level of mercury is 0.5 mg/kg.
1.0 mg/kg, because of historical mercury contamination (Åkerblom Individual susceptibility can also play a role in mercury accu-
and Johansson, 2008). This exposure was not reflected in the mulation. In fact, inter-individual variation in mercury biomarkers
Swedish data from the DEMOCOPHES study probably because may be partly explained by genetic variability. This may explain
consumption of FRESHWATER FISH is small and concentrated to local some of the exceptional cases that were identified in the present
populations, which were not captured in the DEMOCOPHES material. For example one mother in the L1 cluster reported that
sample. she “almost never” consumed fish or marine products. She re-
In the DEMOCOPHES material, the group H3, with 37% of the ported that she ate fish and shellfish “once per month or less” and
studied sample, reported consumption of fish and marine products “almost never” other SEAFOOD PRODUCTS. As mentioned, a “never” an-
once a week or more. The mercury levels in this group are below
swer in the questionnaire could be considered quite reliable. She
1.65 μg/g for 90% of the mothers. When analyzing the contribution
had 2.8 mg/g mercury in the hair which is the range of high SEA FISH
of the different food items, we found that 90% of those who ate
consumers. These exceptions in single cases could be associated
fish but no other marine products had mercury values below
with individual susceptibility or a genetic polymorphism (Basu
0.55 μg/g, and half of the group (P50) had values below 0.15 μg/g.
et al., 2014; Julvez and Grandjean, 2013).
Furthermore, 50% of the mothers consuming SEA FISH several times
Mercury contamination of the human food chain is a significant
per week had mercury values below 0.81 μg/g (Table 2). This
problem for human health and wellbeing. Fish and shellfish are
is valuable information for public health authorities when
important components in a healthy diet and have a high cultural
developing dietary recommendations. For example, the US-EPA
recommended level for women in childbearing age is 1 μg mercury and social value in many parts of Europe. They are central in-
per gram hair (US-EPA, 2001) which means that the whole group gredients in the so-called “Mediterranean diet”, which today is
studied here (eating only SEA FISH once per week independently of recommended to mitigate obesity and cardio-vascular disease.
the species and size) and half of the group of frequent consumers Unfortunately these products can be a significant source of mer-
would be on the “safe side”. Furthermore, if we consider the JECFA/ cury exposure. There is a risk that people stop eating fish and
WHO recommended levels of 1.9 μg/g, 89% of the higher con- shellfish because of the perceived health risks of mercury, which
sumers (eating only SEA FISH) fall below this limit (Table 2). would be highly undesirable from a public health point of view.
A very clear, additional pattern comes out from the analysis. The first recommendation is to focus the advice on the vulnerable
SEAFOOD PRODUCTS, and to a lesser extent SHELLFISH, contributes sig- groups, which in the case of mercury, is pregnant women. Well-
nificantly to the mercury exposure, both in mothers and their designed communications strategies are therefore important in
children. The SEAFOOD PRODUCTS group includes algae, seaweed and order to correctly communicate both the risks and benefits to
surimi that are quite common in “modern diets”. Surimi is a food human health. An example of this comes from the Bermuda
 A. Castaño et al. / Environmental Research 141 (2015) 58–68 67

Islands where a successful communication strategy and public Acknowledgments

health interventions resulted in an 80% reduction of blood mer-
cury levels in a population of pregnant women following the We would like to thanks DEMOCOPHES volunteers for their
provision of dietary advice of local versus global market fish intake participation.
(Dewailly et al., 2012).
It is important to reduce anthropogenic mercury emissions as
far as possible. The ratification of the Minamata Convention will be Appendix A. Supplementary material
an important step in this direction. These kinds of strategies be-
come useless if it is not possible to show that they work. Human Supplementary data associated with this article can be found in
biomonitoring, based on analysis of mercury in hair, is relatively the online version at http://dx.doi.org/10.1016/j.envres.2014.10.
cheap and a very sensitive tool to follow the implementation of 029.
this strategy and determine its effectiveness. It will be able to
clearly demonstrate whether the strategy works and human ex-
posure to mercury is reduced or not. References

Åkerblom, S., Johansson, K., 2008. Kvicksilver i svensk insjö fisk – variationer i tid
och rum. Report 2008/8. Swedish Agricultural University, Institute for En-
5. Conclusions vironmental Analysis (in Swedish).
Basu, N., Goodrich, J.M., Head, J., 2014. Ecogenetics of mercury: From genetic
polymorphisms and epigenetics to risk assessment and decision-making. En-
There is a widespread difference in mercury exposure in the
viron. Toxicol. Chem. 33, 1248–1258.
European population and the difference is very likely related to Becker, K., Seiwert, M., Casteleyn, L., Joas, R., Joas, A., Biot, P., Aerts, D., Castaño, A.,
dietary habits and in particular to consumption of fish and other Esteban, M., Angerer, J., Koch, H.M., Schoeters, G., Hond, E.D., Sepai, O., Exley, K.,
Knudsen, L.E., Horvat, M., Bloemen, L., Kolossa-Gehring, M., 2014. A systematic
products from the marine environment. The frequency of fish and
approach for designing a HBM pilot study for Europe. Int. J. Hyg. Environ.
aquatic products consumption shows a clear pattern across Europe Health 217, 312–322.
with a relatively higher frequency in the Mediterranean and North Casteleyn, L., Dumez, B., Becker, K., Kolossa-Gehring, M., Den Hond, E., Schoeters, G.,
Castaño, A., Koch, H.M., Angerer, J., Esteban, M., Exley, K., Sepai, O., Bloemen, L.,
European countries and a lower frequency in Central European Horvat, M., Knudsen, L.E., Joas, A., Joas, R., Biot, P., Koppen, G., Dewolf, M.-C.,
countries. The hair mercury levels also show the same general Katsonouri, A., Hadjipanayis, A., Cerna, M., Krskova, A., Schwedler, G., Fiddicke,
pattern. The mercury levels in the mothers and their children are U., Nielsen, J.K.S., Jensen, J.F., Rudnai, P., Kozepesy, S., Mulcahy, M., Mannion, R.,
Gutleb, A.C., Fischer, M.E., Ligocka, D., Jakubowski, M., Reis, M.F., Namorado, S.,
strongly correlated indicating a common source of exposure, most Lupsa, I.-R., Gurzau, A.E., Halzlova, K., Jajcaj, M., Mazej, D., Tratnik Snoj, J., Po-
likely the diet. There was no significant contribution from dental sada, M., Lopez, E., Berglund, M., Larsson, K., Lehmann, A., Crettaz, P., Aerts, D.,
amalgams to hair mercury levels. The mercury exposure is related 2015. A pilot study on the feasibility of European harmonized human biomo-
nitoring: challenges and opportunities. Environ. Res. 141, 2–13.
to consumption of marine fish and seafood and the contribution Choi, A.L., Mogensen, U.B., Bjerve, K.S., Debes, F., Weihe, P., Grandjean, P., Budtz-
from freshwater fish was minor in the studied sample. The ma- Jørgensen, E., 2014. Negative confounding by essential fatty acids in methyl-
jority (a 95%) of those consuming fish once a week or more have mercury neurotoxicity associations. Neurotoxicol. Teratol. 42, 85–92.
Den Hond, E., Govarts, E., Willems, H., Smolders, R., Casteleyn, L., Kolossa-Gehring,
mercury levels (0.55 mg/g hair) well below the health based M., Schwedler, G., Seiwert, M., Fiddicke, U., Castaño, A., Esteban, M., Angerer, J.,
limit values recommended by US-EPA (1 mg/g hair) and by WHO Koch, H.M., Schindler, B.K., Sepai, O., Exley, K., Bloemen, L., Horvat, M., Knudsen,
(1.9 mg/g hair) for the most vulnerable population group. Seafood L.B., Joas, A., Joas, R., Biot, P., Aerts, D., Koppen, G., Katsonouri, A., Hadjipanayis,
A., Krskova, A., Maly, M., Mørck, T.A., Rudnai, P., Kozepesy, S., Mulcahy, M.,
products and shellfish were found to significantly contribute to Mannion, R., Gutleb, A.C., Fischer, M.E., Ligocka, D., Jakubowski, M., Reis, M.F.,
mercury exposure and this potential exposure source should be Namorado, S., Gurzau, A.E., Lupsa, I.R., Halzlova, K., Jajcaj, M., Mazej, D., Tratnik,
further monitored. As mercury will be present in the environment J.S., López, A., López, E., Berglund, M., Larsson, K., Lehmann, A., Crettaz, P.,
Schoeters, G., 2015. First steps toward harmonised human biomonitoring in
for many years to come, human biomonitoring programs, like Europe: demonstration project to perform human biomonitoring on a Eur-
DEMOCOPHES, are important tools in assessing current population opean scale. Environ. Health Perspect 123, 255–263.
exposure and in discovering trends and patterns related to mer- Dewailly, E., Rouja, P., Forde, M., Peek-Ball, C., Côté, S., Smith, E., Drescher, O., Ro-
bertson, L., 2012. Evaluation of a public health intervention to lower mercury
cury mitigation policies (Minamata convention), life style and food exposure from fish consumption in Bermuda. PLoS ONE 7, e47388.
consumption. This information is essential for assessing the ef- Esteban, M., Schindler, B.K., Jiménez-Guerrero, J.A., Koch, H.M., Angerer, J., Rivas, T.
C., Rosado, M., Gómez, S., Casteleyn, L., Kolossa-Gehring, M., Becker, K., Scho-
fectiveness of policies and for advisory authorities in developing
eters, G., Den Hond, E., Bloemen, L., Sepai, O., Exley, K., Knudsen, L.E., Horvat,
relevant consumer recommendations with respect to products M., Joas, A., Joas, R., Aerts, D., Biot, P., Borošová, D., Davidson, F., Dumitrascu, I.,
from the aquatic environment. Fischer, M.E., Grander, M., Janasik, B., Jones, K., Kašparová, L., Larssen, T., Naray,
M., Nielsen, F., Hohenblum, P., Pinto, R., Pirard, C., Plateel, G., Tratnik, J.S.,
Wittsiepe, J., 2015. EQUAS reference laboratories, In: Castaño A. Mercury ana-
lysis in hair: Comparability and quality assessment within the transnational
COPHES/ DEMOCOPHES project. Environ. Res. 141, 23–29.
Funding EUMOFA database. 〈http://ec.europa.eu/fisheries/market-observatory/home〉 (ac
cessed 10.07.14).
The DEMOCOPHES Project (LIFE09 ENV/BE/000410) is jointly EUROSTAT. 2014. Fishery statistics. 〈http://epp.eurostat.ec.europa.eu/statistics_ex
plained/index.php/Fishery_statistics〉 (accessed 14.03.14).
funded by the European Commission programme LIFE þ (50%), EFSA, 2012. Scientific opinion on the risk for public health related to the presence of
with the remaining 50% being provided by the participating mercury and methylmercury in food. EFSA J 10, 2985–3136.
countries (see the national implementation websites accessible via Fiddicke, U., Becker, K., Schwedler, G., Seiwert, S., Joas, R., Joas, A., Biot, P., Aerts, D.,
Casteleyn, L., Castaño, A., Esteban, M., Angerer, J., Koch, H.M., Schoeters, G., Den
http://www.eu-hbm.info/democophes/project-partners). The CO- Hond, E., Sepai, O., Exley, K., Knudsen, L.E., Horvat, M., Bloemen, L., Katsonouri,
PHES Project that provided the operational and scientific frame- A., Hadjipanayis, A., Cerna, M., Krskova, A., Mørck, T.A., Rudnai, P., Kozepesy, S.,
work was funded by the European Commission Seventh Frame- Mulcahy, M., Mannion, R., Gutleb, A.C., Fischer, M., Ligocka, D., Jakubowski, M.,
Namorado, S., Reis, M.F., Lupsa, I.R., Gurzau, A.E., Halzlova, K., Jajcaj, M., Mazej,
work Programme-DG Research (Grant agreement no. 244237- D., Tratnikj, J.S., Rivas, T.S., Gómez, S., Berglund, M., Larsson, K., Lehmann, A.,
www.eu-hbm.info). This work would not have been possible Crettaz, P., Kolossa-Gehring, M., 2015. Lessons learnt on recruitment and
without the additional funding from the Spanish Ministry of fieldwork from a pilot European human biomonitoring survey. Environ. Res.
141, 14–22.
Agriculture, Food and Environment (MAGRAMA) and The Institute Frery, N., Saoudi, A., Garnier, R., Zeghnoun, A., Falq, G., Guldner, L., 2010. Exposure
of Health Carlos III (ISCIII) Agreement, SEG 1112/10. of the French Population to Environmental Pollutants – Environmental
68 A. Castaño et al. / Environmental Research 141 (2015) 58–68

Components of the French National Survey on Nutrition and Health – Initial Miklavčič, A., Casetta, A., Snoj Tratnik, J., Mazej, D., Krsnik, M., Mariuz, M., Sofianou,
Results. French Institute for Public Health Surveillance, Saint-Maurice (Fra). K., Spirić, Z., Barbone, F., Horvat, M., 2013. Mercury, arsenic and selenium ex-
Harkins, D., Susten, A.S., 2003. Hair analysis: exploring the state of the science. posure levels in relation to fish consumption in the Mediterranean area. En-
Environ. Health. Perspect. 111, 576–578. viron. Res. 120, 7–17.
Joas, R., Casteleyn, L., Biot, P., Kolossa-Gehring, M., Castano, A., Angerer, J., Schoeters, Myers, G.J., Davidson, P.W., Cox, C., Shamlaye, C., Palumbo, D., Cernichiari, E.,
G., Sepai, O., Knudsen, L.E., Joas, A., Horvat, M., Bloemen, L., 2012. Harmonised Sloane-Reeves, J., Wilding, G.E., Kost, J., Haung, Li-S., Clarkson, T.W., 2003.
human biomonitoring in Europe: activities towards an EU HBM framework. Int. Prenatal methylmercury exposure from ocean fish consumption in the Sey-
J. Hyg. Environ. Health. 215, 172–175. chelles child development study. Lancet 361, 1686–1692.
Johnson, R.A., Wichern, D.W., 1988. Applied Multivariate Statistical Analysis, Second NRC (National Research Council), 2000. Toxicological Effects of Methylmercury.
Edition Prentice Hall International Inc., New Jersey, USA. National Academy Press, Washington, DC.
Julshamn, K., Andersen., A., Ringdal, O., Mørkøre, J., 1987. Trace elements intake in Schoeman, K., Bend, J.R., Hill, J., Nash, K., Koren, G., 2009. Defining a lowest ob-
the Faroe Islands. I. Element levels in edible parts of pilot whales (Globicephalus servable adverse effect hair concentrations of mercury for neurodevelopmental
meleanus). Sci. Total Environ. 65, 53–62. effects of prenatal methylmercury exposure through maternal fish consump-
Julvez, J., Grandjean, P., 2013. Genetic susceptibility to methylmercury develop- tion: a systematic review. Ther. Drug Monit. 31, 670–682.
mental neurotoxicity matters. Front. Genet. 4, 278. Selin, N.E., 2009. Global biogeochemical cycling of mercury: a review. Annu. Rev.
Karagas, M.R., Choi, A.L., Oken, E., Horvat, M., Schoeny, R., Kamai, E., Cowell, W., Environ. Resour. 34, 43–63.
Grandjean, P., Korrick, S., 2012. Evidence on the human health effects of low- Sherman, L.S., Blum, J.D., Franzblau, A., Basu, N., 2013. New insight into biomarkers
level methylmercury exposure. Environ. Health Perspect. 120, 799–806. of human mercury exposure using naturally occurring mercury stable isotopes.
Kjellstrom, T., Kennedy, P., Wallis, S., Mantell, C., 1986. Physical and mental devel- Environ. Sci. Technol. 47, 3403–3409.
opment of children with prenatal exposure to mercury from fish. Stage 1 pre- Tina, N., Nurul, H., Ruzita, A., 2010. Surimi-like material: challenges and prospects.
liminary tests at age 4, Solna, National Swedish Environmental Board. (Report Int. Food Res. J. 17, 509–517.
number 3080). US EPA, 2001. Water Quality Criterion for the Protection of Human Health: Me-
Kurland, L.T., Faro, S.N., Siedler, H., 1960. Minamata disease. The outbreak of a thylmercury. US Environmental Protection Agency.
neurologic disorder in Minamata, Japan, and its relationship to the ingestion of Valent, F., Mariuz, M., Bin, M., Little, D., Mazej, D., Tognin, V., Tratnik, J., McAfee, A.J.,
seafood contaminated by mercuric compounds. World Neurol. 1, 370–395. Mulhern, M.S., Parpinel, M., Carrozzi, M., Horvat, M., Tamburlini, G., Barbone, F.,
Lamborg, C.H., Hammerschmidt, C.R., Bowman, K.L., Swarr, G.J., Munson, K.M., 2013. Associations of prenatal mercury exposure from maternal fish con-
Ohnemus, D.C., Lam, P.J., Heimbürger, L.E., Rijkenberg, M.J.A., Saito, M.A., 2014. sumption and polyunsaturated fatty acids with child neurodevelopment: a
A global ocean inventory of anthropogenic mercury based on water column prospective cohort study in Italy. J. Epidemiol. 23, 360–370.
measurements. Nature 512, 65–68. WHO, 1996. Biological Monitoring of Chemical Exposure in the Workplace, vol. 1.
Mahaffey, K.R., 2004. Fish and shellfish as dietary sources of methylmercury and World Health Organization.
the omega-3 fatty acids, eicosahexaenoic acid and docosahexaenoic acid: risks Yorifuji, T., Tsuda, Toshihide, Harada, M., 2013. Minamata disease: a challenge for
and benefits. Environ. Res. 95, 414–428. democracy and justice, Late Lessons from Early Warnings: Science, Precaution,
Mason, R.P., Choi, A.L., Fitzgerald, W.F., Hammerschmidt, C.R., Lamborg, C.H., Soer- Innovation. European Environment Agency, Copenhagen, Denmark p. 92,
ensen, A.L., Sunderland, E.M., 2012. Mercury biogeochemical cycling in the http://www.eea.europa.eu/publications/late-lessons-2 (accessed 12.03.14).
ocean and policy implication. Environ. Res. 119, 101–107.