Content uploaded by Donna Mergler
Author content
All content in this area was uploaded by Donna Mergler on Feb 01, 2024
Content may be subject to copyright.
The Contribution across Three Generations of Mercury Exposure to Attempted
Suicide among Children and Youth in Grassy Narrows First Nation, Canada: An
Intergenerational Analysis
Donna Mergler,
1
Aline Philibert,
1
Myriam Fillion,
2,1
and Judy Da Silva
3
1
Centre de recherche interdisciplinaire sur le bien-être, la santé, la société et l’environnement, Université du Québec à Montréal, Montréal, Québec, Canada
2
Département Science et Technologie, Université T
ELUQ, Montréal, Québec, Canada
3
Grassy Narrows First Nation, Grassy Narrows, Ontario, Canada
BACKGROUND:For 60 y, the people of Asubpeeschoseewagong Anishinabek (Grassy Narrows First Nation) have endured the effects of massive mer-
cury (Hg) contamination of their river system, central to their traditions, culture, livelihood, and diet. In the years following the Hg discharge into the
English–Wabigoon River system by a chloralkali plant in the early 1970s, there was a dramatic increase in youth suicides. Several authors attributed
this increase solely to social disruption caused by the disaster.
OBJECTIVE:This research examined the possible contribution of Hg exposure across three generations on attempted suicides among today’s children
(5–11 y old) and youth (12–17 y old), using a matrilineal intergenerational paradigm.
METHODS:Information from the 2016–2017 Grassy Narrows Community Health Assessment (GN-CHA) survey was merged with Hg biomonitoring
data from government surveillance programs (1970–1997). Data from 162 children/youth (5–17 years of age), whose mothers (n= 80) had provided
information on themselves, their parents, and children, were retained for analyses. Direct and indirect indicators of Hg exposure included a)grand-
father had worked as a fishing guide, and b)mother’s measured and estimated umbilical cord blood and childhood hair Hg and her fish consump-
tion during pregnancy with this child. Structural equation modeling (SEM) was used to examine significant links from grandparents (G0) to mothers’
exposure and mental health (G1) and children/youth (G2) risk for attempted suicide.
RESULTS:Mothers’(G1) median age was 33 y, 86.3% of grandmothers (G0) had lived in Grassy Narrows territory during their pregnancy, and 52.5%
of grandfathers (G0) had worked as fishing guides. Sixty percent of children (G2) were <12 years of age. Mothers reported that among teenagers
(G2: 12–17 years of age), 41.2% of girls and 10.7% of boys had ever attempted suicide. The SEM suggested two pathways that significantly linked
grandparents (G0) to children’s (G2) attempted suicides: a) through mothers’(G1) prenatal and childhood Hg exposure and psychological distress,
and b) through maternal fish consumption during pregnancy (G1/G2), which is an important contributor to children’s emotional state and behavior.
DISCUSSION:Despite minimal individual information on G0 and G1 past life experiences, the findings support the hypothesis that Hg exposure over
three generations contributes to the mental health of today’s children and youth. The prevalence of Grassy Narrows youth ever having attempted sui-
cide is three times that of other First Nations in Canada. https://doi.org/10.1289/EHP11301
Introduction
Since the first public evidence, 52 y ago, of the extent of the mercury
(Hg) discharge into the English–Wabigoon River system by a chlor-
alkali plant in Dryden, Ontario (Canada), much has been written
about the Asubpeeschoseewagong Anishinabek (Grassy Narrows
First Nation) and how the contamination of their waterways dis-
rupted their culture, traditions, and economy.
1–8
For generations, fish
provided people of Grassy Narrows with “meaning, social cohesion,
and pride.”
9
In the early 1960s, a large majority were gainfully
employed in fish-based industries.
1–9
Approximately 80% of the
households in Grassy Narrows had a least one family member who
worked as a fishing guide and many were involved in commercial
fishing; employment was 85%–90%.
1–3
During the tourist fishing
season, all Grassy Narrows fishing guides reported eating fish every
day as part of the customary shore meal for anglers, and almost all
(92%) brought fish home to their family.
1
In 1970, very high Hg levels were reported in fish in the English–
Wabigoon River system; inorganic Hg released by the chloralkali
plant was being converted into methylmercury (MeHg), a highly
toxic compound that bioaccumulates and biomagnifies in the aquatic
food chain.
5,10
Hg concentrations in fish reached up to 24 lg=g,
almost 50 times the upper limit considered safe for human consump-
tion.
10
Communities received contradictory messages about whether
or not they should continue to eat fish. Some lodge owners down-
played the importance of Hg contamination and continued the prac-
tice of shore meals, thereby placing fishing guides into the situation
of choosing between their health and their livelihood.
1
In 1975, the
main fishing lodge had shut down and the commercial fisheries were
closed.
1
Over time, fish Hg concentrations declined, stabilizing by
around 1985,
10,11
but they have remained to this day higher than in
other Ontario lakes and rivers.
12
In the early 1970s, Canadian government biomonitoring pro-
grams for total Hg in hair, blood, and umbilical cord blood were
initiated in the First Nation communities affected by the dis-
charge into the English–Wabigoon River system.
13,14
In 1976,
annual average individual peak blood total Hg was 23:8lg=L,
with concentrations ranging from 1:5 to 322:9lg=L.
13
Fishing
guides and their families were identified as the groups the most at
risk; the highest blood total Hg concentration (660 lg=L) among
Indigenous communities in Canada was reported for a Grassy
Narrows fishing guide.
15
The biomonitoring programs continued
into the 1990s, when average Hg biomarker concentrations
declined below the Canadian Hg guidelines, mirroring the decline
in fish Hg concentrations.
13,16
The years following 1970 saw the loss of livelihood and were
fraught with social upheaval.
1–5
There was an 8-fold increase of
violent deaths from pre- to post-1970 and the prevalence of alco-
holism was high.
1–4
Several authors have attributed the social crisis
to the loss of traditional lifestyle and the destruction of the very
foundation of their society.
1,2,4–6
In Grassy Narrows First Nation,
the trauma of cultural dispossession and loss of livelihood was
compounded by Hg poisoning from fish consumption. However,
the possible direct contribution of high Hg exposure to the
Address correspondence to Donna Mergler. Email: mergler.donna@uqam.ca
Supplemental Material is available online(https://doi.org/10.1289/EHP11301).
The authors declare no conflict of interest.
Received 24 March 2022; Revised 26 January 2023; Accepted 4 May 2023;
Published 19 July 2023.
Note to readers with disabilities: EHP strives to ensure that all journal
content is accessible to all readers. However, some figures and Supplemental
Material published in EHP articles may not conform to 508 standards due to
the complexity of the information being presented. If you need assistance
accessing journal content, please contact ehpsubmissions@niehs.nih.gov. Our
staffwill work with you to assess and meet your accessibility needs within 3
working days.
Environmental Health Perspectives 077001-1 131(7) July 2023
A Section 508–conformant HTML version of this article
is available at https://doi.org/10.1289/EHP11301.
Research
psychopathology was not considered, despite reports of psychiatric
disorders among patients with organic Hg poisoning. As early as
1971, Eyl
17
noted that the symptoms of organic Hg poisoning
included a lack of interest in home and work, emotional instability
with fits of anger, depression, or rage. In 1968, Tatetsu (cited in
Ekino et al. 2007
18
) described the psychiatric profile of many
patients with Minamata disease, who had been exposed to MeHg
from the consumption of fish and shellfish, contaminated by the
wastewater discharge from a chemical plant into the Minamata
Bay: “They were egocentric, selfish, and paid little attention to the
advice of others. Loss of inhibition was another feature of the dis-
ease’s personality characteristics. They were often impolite, too
friendly to others, easily burst into anger, restless, euphoric, and
childish. Some of them spent money as much as they were given
and threw away change.”
Prior to 1970, no suicide had ever been recorded in Grassy
Narrows.
2
Between 1974 and 1978, three of the four persons who
took their life were teenagers, and in 1977–1978, 26 young persons
between 11 and 19 years of age attempted suicide.
4
Between 1970
and 1977, the incidence of suicide was 3.6 times higher compared
with another First Nation community, selected for its pre-1970
socioeconomic similarities to Grassy Narrows.
1
Neurological
examinations, carried out in Grassy Narrows in the 1970s, focused
primarily on motor and sensory functions, with results that
appeared to be “elusive,”although there were several signs and
symptoms similar to those reported for patients with Minamata dis-
ease.
15
In 2010, Harada et al. examined 73 persons from Grassy
Narrows using the protocol for the diagnosis of Minamata disease
in Japan: 42 (57.4%) were diagnosed with Minamata disease or
suspected Minamata disease, 15 (20.5%) exhibited emotional dis-
turbances, and 10 (13.6%) showed intellectual deficits.
19
A recent
study in Grassy Narrows reported a positive association between
umbilical cord blood total Hg concentration and clustered symp-
toms of affect/mood disorders in adults.
20
For the people of Grassy Narrows, the problems associated
with Hg exposure may be compounded by the intergenerational
consequences of the trauma caused by the Indian Residential
School system.
21–24
Children were taken from their communities
and sent to government-financed boarding schools, administered
by Christian churches. Many children from Grassy Narrows
were sent to these schools.
25,26
In 1953, the principal of the
Presbyterian school in Kenora, Ontario, wrote: “We must face
realistically the fact that the only hope for the Canadian Indian is
eventual assimilation into the white race.”
26
Mental health issues,
particularly emotional well-being, are the most frequently identi-
fied intergenerational impacts of residential schools on
Indigenous health and well-being in Canada.
27
Little is known about the consequences of environmental expo-
sures across generations. Depression in parents has been consis-
tently found to be associated with children’s behavioral problems
and lower cognitive/intellectual/academic performance.
28
During
pregnancy, maternal consumption of Hg-laden fish exposes the
developing fetus; MeHg is actively transported across the placental
barrier and umbilical cord blood MeHg is approximately twice that
of maternal blood.
29
Studies of Inuit children in Canada report
increased risk for borderline intellectual disability
30
and adolescent
anxiety
31
in relation to umbilical cord blood total Hg.
By 2016, >90%of the adult population of Grassy Narrows
were born since the beginning of the disaster in 1962
32
; most
were exposed to Hg in utero and their parents experienced the
immediate impact of the disaster. Today’s children and youth
live with this legacy. The present study seeks to explore intergen-
erational effects of Hg contamination in Grassy Narrows on
children’s emotions, behavior and attempted suicide, using a
maternal-lineage transmission paradigm.
Materials and Methods
Study Design
Between December 2016 and March 2017, Grassy Narrows First
Nation carried out a community health assessment, the Grassy
Narrows Community Health Assessment (GN-CHA) survey.
Participants provided information on themselves, their parents,
and their children (0–17 years of age). Information from the
GN-CHA on grandparents (G0), mothers (G1), and their children
(G2) was merged with a database constructed from archived bio-
markers of Hg exposure from a government monitoring program
carried out between 1970 and 1997.
16
GN-CHA Survey
The GN-CHA survey used a systematic house-to-house sampling
strategy on-reserve, and convenience sampling off-reserve. Two
coordinators from Grassy Narrows devised and supervised the
field work, assisted by nine local surveyors. The questionnaire
was web-based and participants could fill it out on their own, or
with, when needed, the assistance of the surveyor.
The GN-CHA adult (≥18 years of age) survey included 266
questions covering the following areas: demographics, education,
generational attendance of residential school, work activities and
income, diet, general health status (including diabetes care), men-
tal health and well-being, disabilities, injuries, health care access,
physical activity, smoking, drinking and drug use, and participa-
tion in community activities. Most questions were taken from the
First Nation Regional Health Survey 2008/2010,
33
providing a
basis for comparison with other First Nation communities in
Canada. Further questions addressed local fish consumption at
different periods, as well as specific illnesses and symptoms that
have been associated with Hg exposure. The large majority of
adult participants filled it out on their own. Those who did not fill
it out on their own were older persons for whom English was not
their first language. Translation in Anishinabek was available.
Similar to the adult survey, the GN-CHA survey for children
and youth included the same questions as the First Nation Regional
Health Survey 2008/2010,
33
as well as questions on local fish con-
sumption during mother’s pregnancy and the child’s current fish
consumption (174 questions). Parents or caregivers provided the
information on children. In some instances, the youth was present
when the questionnaire was being filled out; some youth ≥15 years
of age filled out the child questionnaire on their own.
On-reserve, 83.6% of the 213 houses were surveyed (302 per-
sons); 89 persons (22.8%) living off-reserve, from 66 households
were likewise surveyed. Data were collected for 391 adult Grassy
Narrows Registered Band members and 353 children (0–17 years
of age). The age and sex distributions of participants were similar
to their distribution in the 2016 Statistics Canada Aboriginal
Household Survey.
32
GN-CHA results were presented to and dis-
cussed with community members during several small group and
community meetings. The final reports were approved by the
Chief and Council and made public on 24 May 2018 for adults
and on 5 December 2018 for children.
For the present study, we retained information provided for
children between 5 and 17 years of age (n= 162) and their mothers
(n= 80). All mothers had filled out the adult and child/youth ques-
tionnaires on their own; 17 youth (26.2%), between 13 and 17 years
of age, were present when their mother filled out the child/youth
questionnaire. The GN-CHA questions retained in the present sta-
tistical analyses are classified by generation as described below.
•Children (G2): Mothers provided information about each
of their children for the following variables: age, sex, physi-
cal and mental health status (on a five-point Likert scale),
current fish consumption, and diagnosed nervous system
Environmental Health Perspectives 077001-2 131(7) July 2023
disorders [yes/no for at least one of the following: cognitive
or mental disabilities, attention-deficit and attention-deficit/
hyperactivity disorder (ADD/ADHD), learning disabilities,
speech/language difficulties, visual problems that cannot be
corrected by glasses, hearing impairment, cerebral palsy,
Bell’s palsy and movement disorders]. Fetal alcohol spec-
trum disorder (FASD) was included in the questionnaire,
but it was excluded from the list of at least one diagnosed
nervous system disorder. Further questions asked whether
in the past 6 months the child had more emotional or
behavioral problems compared with other children of the
same age and sex (yes/no) and whether the child had ever
attempted suicide (yes/no). Children’s school performance
was grouped into three categories: very good/excellent,
good, and fair/unsatisfactory.
Several questions addressed the mother’s pregnancy with
this child. The questions were worded: “If you are the birth
mother, did you have ...?”Gestational diabetes, hyperten-
sion, prematurity and difficulties during childbirth were
grouped into yes/no. Questions likewise asked, “If you are
the birth mother, did you drink a lot or take drugs during
pregnancy?”(yes/no). The child’s prenatal Hg exposure was
based on mother’s reported local fish consumption during her
pregnancy with this child, grouped into three categories (less
than once a month, once a month, at least once a week).
•Mothers (G1): The following variables were retained: age,
education (years of education and having received a high
school diploma), school success (five-point Likert scale
from poor to excellent), having attended a residential school
(yes/no), currently working (yes/no), reasons for not work-
ing, social support (yes/no), struggle to pay for food over
the past 12 months (at least once a month vs. less), current
smoking (yes/no), current heavy drinking as defined by the
2008/2010 Regional Health Survey
33
as five drinks in one
drinking occasion at least once a month in the past 12
months (yes/no), and obesity categorized using body mass
index ≥30 kg=m2(yes/no). Diagnosed nervous system dis-
orders included at least one of the following conditions
diagnosed by a health professional: ADHD, Alzheimer’s
disease, blindness or visual problems that cannot be cor-
rected by glasses, Bell’s palsy, cerebral palsy, cognitive/
mental disorder, epilepsy, Kennedy’s disease, learning dis-
ability, muscular dystrophy, Parkinsonism, psychological/
nervous disorders, and senile dementia.
Psychological distress was assessed using the Kessler
Psychological Distress Scale (K10)
34
for symptoms of anxiety
or depression experienced in the previous month and scored
onascaleof0–40 (0–5: low distress, 6–19: moderate distress,
and ≥20: high psychological distress).
34
Several GN-CHA
questions were about suicidal ideation and attempted suicide;
here we retained the answers to the following questions: “In
the past 12 months, has a close friend or family member died
by suicide?”(yes/no), “Have you ever thought about commit-
ting suicide?”(yes/no), and “Have you ever attempted
suicide?”(yes/no). “When did the suicide attempt occur?”
[during the past year, as an adult, as an adolescent (aged
12–17 y old), as a child (<12 y old)].
•Grandparents (G0): From the information mothers (G1)
filled out about their parents (G0), the following questions
were retained for the present analyses: “Was your father a
fishing guide when you were a child”? (yes/no) and “Did
(your mother, father, grandmother, grandfather, other family
members) attend a residential school?”(yes/no). Mothers pro-
vided further information about where their mother (G0) lived
when she was pregnant with them and their place of birth.
For all questions, the GN-CHA survey provided the
choices of “Idonotknow”and “Refused.”Although few per-
sons refused to answer, a substantial number of answers were
“I do not know,”which we considered as missing data.
Biomarkers of Hg Exposure
At the request of the Grassy Narrows Chief and Council, the First
Nation and Inuit Health Branch of Indigenous Services, Canada,
provided the authors with individual retrospective Hg biomarker
data, collected by governmental biomonitoring programs for
blood and hair (1970–1997) and umbilical cord blood (1970–
1992).
13,16
Blood Hg concentrations were transformed into hair
equivalent measures using a hair-to-blood Hg ratio of 1:250.
35
A
28-y Hg biomarker database was created using the highest mea-
surement of equivalent hair Hg concentration for each year
sampled (657 persons) and umbilical cord blood Hg concentra-
tion (201 persons).
20
Year-based equivalent hair Hg data were available for 208
individuals of the 391 participants in the GN-CHA (53.2%);
137 had at least one Hg hair sample taken during childhood
(between 5 and 15 years of age). Umbilical cord blood Hg con-
centrations were available for 99 persons, making up (48.5%) of
the GN-CHA participants born at the local hospital within the time
period.
G0/G1 prenatal exposure. For the 80 mothers retained for the
present study, direct umbilical cord blood Hg measurements were
used where available. For the missing umbilical cord blood data,
we estimated Hg concentration from the complete umbilical cord
blood Hg database (n= 201). For those born between 1970 and
1992, the umbilical cord blood Hg–predicted arithmetic mean
for their year of birth was attributed as described in Philibert
et al. (2022).
20
For those born between 1962 and 1970, the pre-
dicted arithmetic mean cord blood Hg in 1970 was used
(65:1lg=L). For women whose mother had spent her preg-
nancy and delivered elsewhere, umbilical cord blood Hg was
setat1:0lg=L, which corresponded to the lowest 2.5th percen-
tile of measured cord blood values. The correlation between
measured and estimated values for umbilical cord blood Hg
concentrations was rho = 0:22; p=0:023.
20
G1 childhood Hg exposure. Direct hair Hg concentration
measurements, taken between 5 and 15 years of age, were avail-
able for 39 mothers. For the remaining mothers, childhood esti-
mated Hg was derived from the overall arithmetic mean hair Hg
in the year in which they turned 10 y old and then adjusted on the
basis of reported child fish consumption and having attended a
residential school, as previously described.
20
The correlation
between measured and estimated childhood hair Hg concentra-
tions was rho = 0:73; p<0:001.
20
No Hg measurements were available for the G0 or G2 genera-
tions. For the former, the G1 answer to the question: “Was your
father a fishing guide when you were a child?”was used as an in-
dicator of higher Hg exposure, and for children (G2), mother’s
fish consumption during this pregnancy (G1/G2) and child’s cur-
rent fish consumption, were used.
Statistical Analyses
Complex networks with causal pathways between variables are
the essence of intergenerational analyses. Structural equation
modeling (SEM) provides an appropriate framework for illus-
trating pathways and testing statistical associations. SEM allows
for simultaneous consideration of multiple exposures and out-
comes
36
while providing direct and indirect relations and flexi-
bility in modeling covariances.
37–40
The SEM path diagrams
Environmental Health Perspectives 077001-3 131(7) July 2023
illustrate and quantify the contribution of variables and trans-
mission of effects.
Using SEMs, we tested the pathways relevant to Hg exposure
between G0, G1, and G2 with respect to the final outcomes of
children and youth’s emotional status and behavior and attempted
suicide. The correlations between error terms of the different var-
iables were verified and taken into account in the models. The
pathways used a time order based on the premise of chronologi-
cal occurrences (causes precede their effects). To control for con-
founding, we ensured that all backdoor paths on the SEM were
closed (adjustment of covariates). Covariates along pathways
were kept in the model at a p≤0:10 or if they substantively
altered the model (≥20%change). Based on underlying assump-
tions, mediation (i.e., the chain of events) was tested by examin-
ing direct and indirect effects. The moderating effects of mothers’
age or age at birth were tested.
Model goodness of fit was verified using the comparative fit
index (CFI),
39,41
Tucker–Lewis index (TLI),
40
and global fit index
(GFI),
38
as well as chi-square divided by the degrees of freedom
(v2=df), root mean square error of approximation (RMSEA),
39,41
standardized root mean square residual (SRMR),
42
and Schwarz’s
Bayesian information criterion (BIC).
39,41,43
To improve goodness
of fit, we tested the addition/deletion of variables, using modifica-
tion indices (MIs).
39
The sequence of initial fit, modification, and
refitting were repeated until we obtained the SEM with the best fit.
To support the SEM, in parallel, a series of directed acyclic
graphs (DAGs) were performed to confirm the structure of rela-
tions between variables and to prevent misspecification or misin-
terpretation from possible biases. DAGs are useful for representing
conditional independency among variables by evaluating data con-
sistency and for controlling backdoor paths with minimal sufficient
adjustment sets.
43,44
To support the nature and size of the relation-
ship between two variables along the SEM pathways, a series of
logistic regression models were used to examine the direct associa-
tions and provide an indication of the magnitude of effects.
Given the potential for errors associated with creating Hg ex-
posure estimates from data based on surveillance programs with
no precise sampling strategy,
13
as well as the inherent variance in
blood to hair conversion
45,46
where possible, we examined the
associations between measured Hg concentrations and reported
fish consumption. We reran the SEM using reported G1 fish con-
sumption during childhood in the place of estimated Hg exposure
during childhood.
To ensure that the database of 80 mothers and their 162 chil-
dren was sufficient to run SEM, two types of power analyses
were performed: a) to detect model misspecification, and b)to
detect target effects between associations.
39,41,43
Post hoc power
analyses were run to detect model misspecification based on
likelihood-ratio (LR) chi-square and RMSEA tests of close and
not-close fit.
47–51
The power for detecting specific target effect
for direct and indirect pathways was determined with simulated
data in SEM for 100 and 500 simulations on the 162 children.
51
Because some children shared the same mother, we tested the
nonindependent sample using clustering (mother), with the lav-
aan.survey R package.
52
To reduce bias introduced by missing data, the full-information
maximum likelihood (FIML) technique in the lavaan R package
was used to handle missing data.
39,53
Even though several
authors
54,55
consider that logit FIML can accommodate missing
categorical data with unbiased errors and fit parameters, we veri-
fied the model with the weighted least squares (WLS) mean- and
variance-adjusted chi-square test of model fit (WLSMV) estima-
tor, which handles nonnormal and categorical variables, based on
complete cases after multiple imputation.
39,56
When using FIML, we
ran the maximum likelihood with robust standard errors (MLR), as
well as the maximum likelihood estimation with Huber–White
standard errors (Hubert–White sandwich), because it supports com-
plete and incomplete data.
39
Hg estimates for umbilical cord blood and childhood hair con-
centrations were calculated from exposure prediction models,
20
which included averaged year-based measured Hg exposure. Our
assumption was that true Hg exposure varied randomly around
the year-based average (Berkson type error). The Berkson type
error would have the effect of lowering the power of analyses.
57
To control for this, we corrected with bootstrap confidence inter-
vals (CIs) for relative risks.
Bootstrap standard errors and p-values analyses were carried
out to increase SEM accuracy and its reliability (lavaan R pack-
age). The Bollen.stine technique, where data is first transformed
such that the null hypothesis holds exactly in the resampling
space, was verified to extract fit parameters from a fitted lavaan
object.
58,59
We used the false discovery rate (FDR) measure for
controlling type I error of multiple comparisons.
60
We compared the performance of SEM by comparing the fit
parameters, point estimates, and standard errors to the complex
sampling design (cluster-based), which takes into account mul-
tiple children. In the case of small differences between non
clustered-based and cluster-based analyses, we preferred to
use the conventional analysis using FIML
53
rather than data
imputation as required by the mother-clustered–based model
(lavaan.survey R package).
39,61
Variables with a skewed distribution were log (base 10) trans-
formed. Threshold of statistical significance in all analyses was
set at p≤0:05.
Database management and descriptive statistical analyses
were performed using JMP Professional (version 16.0; Statistical
Analysis Hardware; SAS Institute Inc.). All other analyses used
the R statistical software (version 3.6.1; R Development Core
Team). SEMs were computed with the lavaan R package.
39,62
Further analyses with DAGs used dagitty and ggdag R packages.
Power analyses for model misspecification were computed using
the semPower R package. An online tool
51
was used to assess
power for detecting target effects. The clustering (mother) effect
was tested with the lavaan.survey R package along with mice and
mitools R packages for data imputation.
Results
The characteristics reported by mothers (n= 80) for themselves
(G1) and their parents (G0) are presented in Table 1. The large
majority of grandmothers (G0) resided in Grassy Narrows terri-
tory during their pregnancy (n= 69), and half of grandfathers
(G0; n= 42) had been fishing guides. Many grandparents (G0)
had attended residential schools; for all mothers (G1) ≥35 years
of age (n= 31), at least one parent (G0) had been taken to a resi-
dential school.
Mothers’(G1; n= 80) median age was 33 y old, ranging from
23 to 55 [interquartile range (IQR): 29–37 y]. Most mothers (G1;
n= 77) had some high school education, but only 21% (n= 17)
had received a high school diploma (Table 1). At the time of the
survey, approximately half of the mothers (G1; n= 41) were
working; those who were not looking for work at the time of the
survey were either an at-home parent, in poor health or disabled,
or a student. The most prevalent nonneurologic chronic health
conditions were allergies (n= 25; 33%), eczema (n= 19; 24%),
arthritis (n= 11; 15%), and diabetes (n= 11; 15%). Diagnosed
neurologic disorders were reported by 14 mothers (Table 1); all
scored in the moderate/high K10 Psychological Distress Scale
compared with 56% (n= 34) of those who reported no diagnosis.
Athirdofmothers(n= 27) reported that a family member or
close friend had committed suicide, and half (n=40) reported
Environmental Health Perspectives 077001-4 131(7) July 2023
that they themselves had thought about suicide (Table 1).
Among those who had ever thought about suicide, 31 (77.5%)
had attempted suicide at least once in their lives; for over half
of these (n= 19; 61.2%), the first suicide attempt occurred dur-
ing childhood/adolescence. More mothers who had thought
about suicide scored in the moderate-high range of the K10
Psychological Distress Scale compared with those who had not
(n=31; 81.6% vs. n= 13; 40.6% LR chi-square = 12:82; Fisher
exact test: p<0:001).
Some mothers (G1) did not know whether their father (G0)
had been a fishing guide or not (n= 14; 17.5%). Comparison of
the characteristics of mothers who reported that their father had
been a fishing guide (n= 42) and those who reported that their
father had not been a fishing guide (n= 24) are presented in
Table S1. No difference was observed with respect to their
parents (G0) having been placed a residential school; however,
all mothers (G0) of the 42 women whose father had been a fish-
ing guide, resided on Grassy Narrows territory during their preg-
nancy, compared with 75% for the mothers of the 24 women
whose father had not been a fishing guide (Table S1).
For the mothers who knew whether their father had been a fish-
ing guide (G1) (n=66), no significant differences were observed
with respect to the father (G0) having been a fishing guide or not
for sociodemographic and health characteristics (Table S1), with
the exception of diagnosed neurologic disorders: 23.8% (n=10 of
42) of those who knew that their father had been a fishing guide
and none of those who knew that he had not been a fishing guide
(Table S1). Childhood fish consumption and suicidal ideation were
higher among the women who reported that their father had been a
fishing guide (n=24;n= 25 of 42, respectively), but the difference
was just above the significance threshold (Table S1). More women
whose father had been a fishing guide scored in the high range of
the K10 Psychological Distress Scale (n= 11; 26.8%) compared
with only one for the others.
Table 2 contains the measured and estimated mean, median,
and IQR values for mothers’umbilical cord blood and hair Hg.
There was no difference in measured umbilical cord Hg between
mothers who reported that their father had been a fishing guide
compared with those who indicated that their father had not been a
fishing guide [median = 8:5lg=L; 75th percentile: 8:7lg=L
(n=17) vs. median=5:7lg=L; 75th percentile: 17:5lg=L
(n= 7); Wilcoxon/Kruskal–Wallis tests (rank sums): S= 108;
p=0:204], but their measured childhood hair Hg was signifi-
cantly hi gher [median = 0:77 lg=g; 75th percentile: 1:1lg=gvs.
median = 0:5lg=g; 75th percentile: 0:7lg=g; Wilcoxon/Kruskal–
Wallis tests (rank sums): S=140; p=0:045], paralleling their
reported fish consumption.
Table 3 contains information, reported by the mothers, for
each pregnancy and the child’s health and well-being (G2;
n= 162). Mothers filled out the questionnaire for an average of
two children, ranging from one to seven children. For almost
20% of children (n= 31), mothers reported eating fish at least
once a week during their pregnancy (Table 3). Mothers’fish
consumption during pregnancy reflected her own childhood fish
consumption [Kendall’s Tau-b: 0.46 (95% CI: 0.33, 0.57)
Fisher’s exact test: p<0:001] and was higher among those
whose father (G0) was a fishing guide [Kendall’sTau-b:0.355
(95% CI: 0.210, 0.510) Fisher’s exact test: p<0:001]. Indeed,
21 mothers (28.4%), whose father (G0) had been a fishing guide
reported eating fish at least once a week or more compared with
only 2 (4.5%) among those whose father had not been a fishing
guide (LR chi-square: 20.78; Fisher exact test: p<0:001). No
associations were observed for mothers’age, maternal age at
pregnancy, or child’s age with respect to G1/G2 fish consump-
tion during pregnancy [Wilcoxon/Kruskal–Wallis tests (rank
sums) one-way test chi-square = 3:39, p=0:188; 2.15, p=0:342;
and 1.97, p=0:374, respectively].
Table 2. Measured and estimated mothers’(G1) umbilical cord blood and
childhood hair mercury (Hg).
nMean Median IQR Min–max
Measured umbilical cord
blood Hg (lg=L)
29 7.65 5.70 2.35–8.80 1.5–35.7
Estimated umbilical cord
blood Hg (lg=L)
51 15.15 5.25 3.59–16.86 1.0–65.2
Measured childhood hair
Hg (lg=g)
39 1.50 0.50 0.50–0.90 0.50–11.4
Estimated childhood hair
Hg (lg=g)
38 1.52 0.92 0.72–2.60 0.50–5.64
Note: IQR, interquartile range; max, maximum; min, minimum.
Table 1. Descriptive characteristics for grandparents (G0) and mothers (G1) as reported by mothers (G1) in the Grassy Narrows Community Health
Assessment (GN-CHA) survey; n= 80.
Characteristics Yes [n(%)] No [n(%)] Missing [n(%)]
a
Generation grandparents (G0)
Grandmother pregnant in GN territory 69 (86.25) 8 (10.00) 3 (3.75)
Grandfather a fishing guide 42 (52.50) 24 (30.00) 14 (17.50)
Grandmother attended a residential school 47 (58.75) 29 (36.25) 4 (5.00)
Grandfather attended a residential school 46 (57.50) 25 (31.25) 9 (11.25)
Generation mother (G1)
Age <35 y 46 (57.50) 34 (42.50) 0
Education (high school diploma) 17 (21.25) 61 (76.25) 2 (2.50)
School success (very good/excellent) 28 (35.00) 49 (61.25) 3 (3.75)
Childhood fish consumption (≥several times a week) 38 (47.50) 36 (45.00) 6 (7.50)
Currently working 41 (51.25) 37 (46.25) 2 (2.50)
Unemployed but looking for work 19 (51.35) 16 (43.24) 2 (5.41)
Struggle to pay for food (at least once a month) 43 (53.75) 27 (33.75) 10 (12.50)
Health perception: thriving (very good/excellent) 17 (21.25) 63 (78.75) 0
At least one chronic health condition 56 (70.00) 23 (28.75) 1 (1.25)
At least one diagnosed nervous system disorder 14 (17.50) 65 (81.25) 1 (1.25)
Obesity ≥30 kg=m244 (55.00) 34 (42.50) 2 (2.50)
Current heavy drinker 44 (55.00) 32 (40.00) 4 (5.00)
Family or close friend has died from suicide last 12 months 27 (33.75) 48 (60.00) 5 (6.25)
Suicidal ideation 40 (50.00) 34 (42.50) 6 (7.50)
Ever attempted suicide 31 (38.75) 43 (53.75) 6 (7.50)
K10 (moderate/high) 49 (61.25) 27 (33.75) 4 (5.00)
a
For each question in the GN-CHA, persons could provide the reply “I don’t know,”“I can’t recall,”or “Refused”, which were grouped as “Missing.”
Environmental Health Perspectives 077001-5 131(7) July 2023
Seventeen mothers (21.3%) reported drinking or taking drugs
for 24 pregnancies (Table 3), representing 15.1% of pregnancies
for which this information was available (n= 158). Mothers who
reported drinking alcohol or taking drugs during pregnancy for
their child (G0/G1) were significantly younger at childbirth com-
pared with those who reported not drinking or taking drugs
(median =19 y; IQR: 18–21 y and median = 23 y; IQR: 20–27 y,
respectively; Wilcoxon/Kruskal–Wallis chi-square: 11.83; Fisher
exact test: p=0:001). The proportion of pregnancies for which the
mother reported drinking or taking drugs was higher for those
whose father (G0) had been a fishing guide compared with those
who reported that their father had not been a fishing guide (n=14,
19.7% vs. n= 2, 4.2%). No association was observed between
drinking or taking drugs during pregnancy and the mother’s current
drinking habits (LR chi-square: 0.06; Fisher exact test: p=0:801).
Few children (n= 4, 2.5%) had a diagnosis of FASD.
GN-CHA mothers’(G1) reports about their children (G2)
are presented in Table 3. Children’s median age was 10 y (IQR:
8–14). For almost one quarter (n= 39), their mothers did not
know if their grandfather had worked as a fishing guide. Most
children (74.7%; n= 121) had not eaten local fish often over the
year prior to the survey (Table 3). The First Nations Regional
Health Survey 2008/2010 defined a measure for conditions that
“may have an impact on learning ability”as at least one of the
following diagnosed conditions: cognitive or mental disabilities,
ADD/ADHD, learning disabilities, speech/language difficulties
33
;
24 children (14.8%) fell into this category. No difference was
observed between boys (n= 14) and girls (n= 10) (LR chi-square:
2.27; Fisher exact test: p=0:132).
Mothers’perceived the mental health of almost 40% (n= 62) of
their children as less than very good or excellent (Table 3), and for
25% of children, the mother reported that the child had more emo-
tional or behavioral problems compared with other children of the
same age and sex (Table 3). The prevalence of emotional or behav-
ioral problems was twice as high in youth (≥12 years of age) com-
pared with younger children (36.7%; n= 24 vs. 18.8%; n= 18; LR
chi-square: 8.87; Fisher exact test: p=0:003). Among younger
children (<12 years of age), no difference was observed between
boys and girls [n= 9 of 41 (22.95%) vs. n= 9 of 48 (18.75%),
respectively; LR chi-square: 4.65; Fisher exact test: p=0:794].
However, more adolescent girls (≥12 years of age) presented emo-
tional or behavioral problems compared with boys [n=17 of 30
(56.67%) vs. n= 7 of 25 (28.00%), respectively; LR chi-square:
4.65; Fisher exact test: p=0:031].
Mothers reported that 13% (n= 22) of children/youth had
attempted suicide, but the percentage rose to 27% (n= 17) when
considering only those ≥12 years of age (Table 3). Among adoles-
cents (≥12 years of age), suicide attempts were higher among girls
(n= 14 of 34; 41.18%) compared with boys (n= 3 of 28; 10.71%;
Fisher exact test LR chi-square: 7.70; p=0:006). None of the chil-
dren with a diagnosis of FASD had attempted suicide.
The SEM in Figure 1 presents the intergenerational pathways
for children and youth attempted suicide, with the corresponding
fit parameters in Table 4 and estimates of sequential mediations
in Table 5. The pathways began with mothers’year of birth,
which represented not only her age, but also accounted for the
change in Hg exposure over time. Different pathways linked G0
through G1 with an increased risk for G2 attempted suicide.
One modeled pathway originated from mothers’umbilical cord
blood Hg (G0/G1), a reflection of (G0) grandmothers’fish consump-
tion, as well as maternal prenatal exposure, and grandfather having
worked as a fishing guide (G0), both of which are seen to contribute
to mothers’childhood Hg (G1) (Figure 1). Mothers’childhood Hg
(G1) was then associated with her being diagnosed with at least one
neurological disorder, which in turn, was associated with current
psychological distress, to which suicidal ideation, over her lifetime,
also contributed. The model further suggested an association
between mothers’K10 psychological distress score (G1) and child’s
mental health status (G2), which was linked to the child’s emotional
state and behavioral problems, which, in turn, may contribute to an
increased risk for attempted suicide, particularly among older girls.
A second modeled pathway lead from the father as a fishing
guide (G0) (Figure 1) to daughters’fish consumption during
pregnancy (G1) and consequently her children’s prenatal Hg ex-
posure (G1/G2). Maternal fish consumption during pregnancy
(G1/G2) contributed to children’s (G2) poor mental health and
emotional and behavioral problems, the latter of which was
directly associated to attempted suicide.
Another pathway showed associations between the grand-
father having been a fishing guide (G0) and mothers’drinking
or taking drugs during pregnancy (G1/G2) (Figure 1). The
association between mothers’drinking and drugs during preg-
nancy and child’s emotional and behavioral problems was not
significant (p=0:14).
For the SEM (Figure 1), power analyses on model misspecifi-
cation showed that a sample size of 162 children was associated
with a power of 71.8%, which was acceptable. Table 4 shows the
SEM fit parameters for the final model and the sensitivity analyses
Table 3. Descriptive characteristics for pregnancy (G1/G2) and children/youth (G2), as reported by mothers (G1) in the Grassy Narrows Community Health
Assessment (GN-CHA) survey: n= 162.
Characteristics Yes [n(%)] No [n(%)] Missing [n(%)]
a
Pregnancy (G1/G2)
Mother’s age at birth of child (≥22 y) 93 (57.41) 69 (42.59) 0
Maternal fish consumption during pregnancy (≥once a week) 31 (19.14) 125 (77.16) 6 (3.70)
Problems during pregnancy/childbirth 33 (20.37) 125 (77.16) 4 (2.47)
Mother’s drinking/drugs during pregnancy 24 (14.81) 134 (82.72) 4 (2.47)
Generation children/youth (G2)
Age (≥12 y) 66 (40.74) 96 (59.26) 0
Sex (girls) 89 (54.94) 73 (45.06) 0
Grandfather was a fishing guide 74 (45.68) 49 (30.24) 39 (24.07)
Fish consumption past year (often) 25 (15.40) 121 (74.69) 16 (9.88)
Overall health (very good/excellent) 129 (79.63) 33 (20.37) 0
Mental health (very good/excellent) 98 (60.49) 62 (38.27) 2 (1.23)
Speech/language disorder 17 (10.49) 139 (85.80) 6 (3.70)
Anxiety or depression 16 (9.88) 134 (82.72) 12 (7.40)
Emotional or behavioral problems 42 (25.93) 102 (62.96) 18 (11.11)
Emotional or behavioral problems (12–17 y) 24 (36.36) 31 (49.97) 11 (16.67)
Ever attempted suicide (all children) 22 (13.58) 133 (82.10) 7 (4.32)
Ever attempted suicide (12–17 y) 17 (27.42) 45 (72.58) 4 (6.45)
a
For each question in the GN-CHA, persons could provide the reply “I don’t know,”“I can’t recall,”or “Refused”, which were grouped as “Missing.”
Environmental Health Perspectives 077001-6 131(7) July 2023
using maternal childhood fish consumption and a mother-clus-
tered–based model. The performance of the SEM models that a)
replaced estimated maternal childhood Hg exposure by her
reported fish consumption at ∼10 years of age (G1), and b) used a
mother-clustered–based model, with imputation, were similar.
Estimates of sequential mediation at different stages in the
SEM pathways are presented in Table 5, with target effects
power analyses simulation for each link. For all pathways to-
ward children and youth (G2) attempted suicide, effects dimin-
ished over generations, but remained significant (Table 5). The
strongest contributions passed through fish consumption during
pregnancy. The pathway that passed through mothers’drinking
or taking drugs during pregnancy (d1 × d2 ×a6 in Figure 1)to
children’s attempted suicide was not significant (coefficient
estimate = 0:01; p=0:17).
Finally, children’s risk of attempting suicide was significantly
influenced by mothers’age at childbirth (Figure 1). The median
age at childbirth for mothers of children who attempted suicide
was 19.5 y (75th percentile: 22 y), whereas the median age at
childbirth of mothers whose child had never attempted suicide
was 23 y (75th percentile: 27 y) (Wilcoxon/Kruskal–Wallis chi-
square: 12.2; p=0:001).
Tables 1 and 3include several characteristics that were tested
in the SEM, but which were not retained in the final model. For
G0, residential school attendance (grandmother and grandfa-
ther) did not enter the model (p>0:1). A total of 59 mothers
(78.7%) reported that at least one of their parents had attended
a residential school; these mothers were significantly older
than those whose parents had not been taken to a residential
school [median = 36 y (IQR: 31–39 y) vs. median = 28:5y
(IQR: 25–30 y); Wilcoxon chi-square = 19 :7; p<0:0001]. For
G1, the following variables were tested but did not enter the model
(p>0:1): education and school success, currently working, strug-
gle to pay for food, heavy drinking, general health, obesity, at least
one chronic health condition, and current fish consumption; for G1/
G2: health issues during pregnancy, prematurity, and difficulties
during childbirth; for G2: physical health, fish consumption over
the past year, and school performance.
Figure 1. Structural equation model pathway diagram on the psychological impact of Hg exposure through a matrilineal lens across three generations.
Intergenerational information, provided by mothers (n= 80) for the Grassy Narrows Community Health Assessment provided for 162 children. Mothers’umbil-
ical cord Hg and childhood hair Hg concentrations were estimated from a 1970–1997 biomarker database. Note: BIC, Schwarz’s Bayesian information crite-
rion; CFI, comparative fit index; Chi2, chi-square; df, degrees of freedom; GFI, global fit index; Hg, mercury; RMSEA, root mean square error of
approximation; SRMR, standardized root mean square residual; TLI, Tucker–Lewis index. p-Values: significance of pathway coefficient estimates:
*
p≤0:05;
**
p≤0:01;
***
p≤0:001.
Table 4. SEM Fit parameters for the final model, the model with maternal childhood fish consumption and the mother-clustered–based model.
SEM model Chi2 df CFI TLI GFI RMSEA (95% CI) SRMR BIC
Final model (Figure 1) 161 80 0.90 0.87 0.99 0.08 (0.06, 0.10) 0.09 3,352
Maternal childhood fish consumption 153 80 0.90 0.88 0.99 0.08 (0.06, 0.10) 0.09 3,793
Mother-clustered 201 80 0.86 0.82 0.99 0.10 (0.08, 0.11) 0.09 3,753
Note: BIC, Schwarz’s Bayesian information criterion; CFI, comparative fit index; Chi2, chi-square; CI, confidence interval; df, degrees of freedom; GFI, global fit index; RMSEA,
root mean square error of approximation; SEM, structural equation model; SRMR, standardized root mean square residual; TLI, Tucker–Lewis index.
Environmental Health Perspectives 077001-7 131(7) July 2023
Table S2 presents a series of multiple regression models for
the various components of the SEM pathways, without mediating
variables. The results support the direct associations computed by
the SEM.
Discussion
This study, which examined a matrilineal linkage over three gen-
erations, suggests an intergenerational impact of Hg exposure on
Grassy Narrows First Nation children’s emotions and behavior
and attempted suicides. Mothers who participated in the GN-
CHA provided information on their parents and children. Most of
these women, born between 1962 and 1993, were exposed to Hg
prenatally and as children, teenagers, and adults. During the years
when Hg exposure was at its highest, their families suffered the
dramatic consequences of the loss of their Indigenous culture and
values, as well as their traditional food and livelihood.
1–6
In the
absence of information on exposure of the grandparent generation
(G0), we used the maternal father’s occupation as a fishing guide
as a proxy for grandfathers’high Hg exposure, based on biomoni-
toring data.
13,14
Maternal umbilical cord blood Hg (G0/G1)
reflected grandmothers’Hg exposure through fish consumption
during pregnancy.
The point of departure for the study was the prevalence of
attempted suicide among today’s children and youth. Mothers
reported that 27.4% of adolescents (n= 17) had attempted suicide
at least once: girls (n= 14; 41.2%) and boys (n= 3; 10.7%).
These prevalences are considerably higher than those reported in
the most recent First Nation and Regional Health Survey,
63
administered to 4,639 First Nation youth at the same period as
the GN-CHA, in which 10.3% reported ever attempting suicide
(15.6% of girls and 5.2% of boys).
In the SEM of the present study, several pathways linked
grandparents (G0) Hg to an increased risk of attempted suicide
for G2 children and youth. The pathway through mothers’(G1)
childhood Hg exposure was mediated by her umbilical cord
blood Hg concentration (G0/G1) and having a father a fishing
guide (G0), both of which likely reflect grandparents fish con-
sumption and Hg exposure. Mothers’(G1) childhood Hg expo-
sure was directly associated with having at least one nervous
system disorder, which, in turn, was associated with her current
psychological distress. These findings are consistent with studies
that have examined Hg exposure and psychological distress.
Using the General Health Questionnaire (GHQ), a study in Japan
reported a high prevalence of psychological distress among 86%
of 133 women ≥40 years of age living in the Minamata area,
where they were exposed to Hg as children.
64
In a recent study of
an Indigenous community in the Brazilian Amazon, hair Hg
among persons 12–72 years of age [mean age 27.4 y (n= 109)]
was associated with depressive symptoms, as assessed with the
Geriatric Depression Scale Short Form; persons with hair Hg
≥10 lg=g were 1.8 times more likely to manifest depressive
symptoms.
65
In the GN-CHA, one mother in five reported a high
level of psychological distress on the Kessler Psychological
Distress Scale. This is consistent with Harada’s observations in
2010 where 20% of 73 Grassy Narrows examinees manifested
signs and symptoms of emotional disturbances.
19
In the present
study, maternal psychological distress score was associated with
child’s poorer health. This association has been observed in
many studies that have linked mothers’psychological distress to
their children’s mental health status.
66,67
Based on the SEM, the strongest contribution of G0 on G2
was through grandfather’sinfluence on mothers’fish consump-
tion during pregnancy, suggesting that the tradition of fish con-
sumption is passed on from father to daughter. Indeed, women
who consumed fish as children, tended to consume fish as adults.
Grassy Narrows elder J.D.S., co-investigator on the present study,
notes that: “fish has always played a vital role in Anishinaabek
natal traditions. The grandmothers used to say, ‘Eat fish broth,
and the breast milk will come out.’” In 2005, Chan et al.
68
reported a very high correlation between local fish consumption
and hair Hg concentrations in Grassy Narrows; for women of
Table 5. Estimates of sequential mediation in SEM pathways for Hg exposure across three generations to attempted suicide in children and youth.
Pathways Estimate (95% CI) p-Value
Power target
simulations
n= 100 n= 500
Toward psychological distress (G1)
Umbilical cord blood Hg (G0/G1) a1 × a2 × a3 0.07 (0.02, 0.12) 0.017 1.00 1.00
Childhood Hg (G1) a2 × a3 0.16 (0.07, 0.25) 0.000 1.00 1.00
Father fishing guide (G0) b1 × a2 × a3 0.04 (0.01, 0.06) 0.000 0.87 0.91
Toward decreased child’s mental health (G2)
Umbilical cord blood Hg (G0/G1) a1 × a2 × a3 × a4 0.02 (0.01, 0.04) 0.017 0.99 0.99
Childhood Hg (G1) a2 × a3 × a4 0.05 (0.02, 0.09) 0.000 0.99 0.99
Fish consumption during pregnancy
(G1/G2)
c2 0.24 (0.09, 0.39) 0.000 0.95 0.96
Father fishing guide (G0) ðb1 × a2 × a3 × a4Þ+ðc1 × c2Þ0.11 (0.03, 0.18) 0.000 0.90 0.92
Toward child’s emotional or behavioral issues (G2)
Umbilical cord blood Hg (G0/G1) a1 × a2 × a3 × a4 × a5 0.01 (0.00, 0.02) 0.000 0.95 0.95
Childhood Hg (G1) a2 × a3 × a4 × a5 0.02 (0.00, 0.04) 0.028 0.96 0.97
Fish consumption during pregnancy
(G1/G2)
ðc2 × a5Þ+ c3 0.33 (0.15, 0.50) 0.000 0.99 0.99
Father fishing guide (G0) ðb1 × a2 × a3 × a4 × a5Þ+ðc1 × c2 × a5Þ
+ðd1 × d2Þ+c3
0.17 (0.07, 0.27) 0.000 0.99 0.99
Toward child’s attempted suicide (G2)
Umbilical cord blood Hg (G0/G1) a1 × a2 × a3 × a4 × a5 × a6 0.00 (0.00, 0.01) 0.010 0.73 0.77
Childhood Hg (G1) a2 × a3 × a4 × a5 × a6 0.01 (0.00, 0.01) 0.044 0.83 0.85
Fish consumption during pregnancy
(G1/G2)
ðc2 × a5 × a6Þ+ðc3 × a6Þ0.11 (0.03, 0.19) 0.017 0.96 0.97
Father fishing guide (G0) ðb1 × a2 × a3 × a4 × a5 × a6Þ+ðc1 × c2 × a5 × a6Þ
+ðd1 × d2 × a6Þ+ðc1 × c3 × a6Þ
0.06 (0.01, 0.10) 0.017 0.97 0.99
Note: Pathways with “a”originated from mothers’umbilical cord blood Hg; Pathways with “b,”“c,”or “d”originated from grandfather a fishing guide. Hg, mercury; SEM, structural
equation modeling. p-Values adjusted on bootstrap: significance of pathway coefficient estimates:
*
p≤0:05;
**
p≤0:01;
***
p≤0:001.
Environmental Health Perspectives 077001-8 131(7) July 2023
child-bearing age (18–40 y), the average fish intake varied by
season and was highest during the summer months ( ∼28 g=d)
and lowest during the winter months ( ∼2g=d). In 2010, Neff
et al.,
10
who analyzed Hg concentrations in fish in the English–
Wabigoon River system, indicated that they had not declined sig-
nificantly since the mid-1980s and may still present a potential
health risk to humans.
In the SEM, maternal fish consumption during pregnancy
(G1/G2), an indicator of prenatal MeHg exposure, was the major
contributor to children’s mental health and emotional state and
behavior, suggesting that at recent fish Hg concentrations, MeHg
was affecting fetal development. MeHg is actively transported
across the placenta and its concentration in umbilical cord blood
has been measured to be ∼1:7 times (95th percentile: 3.4) that of
maternal blood.
29,69,70
Several studies have reported an associa-
tion between prenatal Hg exposure and anxiety-related symptoms
in children or adolescents.
18,31,71–73
These findings are supported
by recent imaging studies of prefrontal brain areas of Inuit ado-
lescents.
74
For the children and youth in the present study, there
was a direct association between emotional state and behavioral
problems and attempted suicide.
The SEM results suggested that despite the absence of spe-
cific information on G0 or about the trauma suffered by mothers
(G1) during their own childhood or adolescence, a pathway
linked grandfather having been a fishing guide to mothers’drink-
ing or taking drugs during pregnancy. The vivid descriptions of
life in Grassy Narrows in the years of the highest Hg exposures
and the socioeconomic and cultural trauma suggest that the con-
text in which these women were raised would be an important de-
terminant of their adolescent and early adult behavior. Negative
childhood experiences have been associated with early alcohol
drinking.
75
In the present study, drinking alcohol or taking drugs
was more prevalent during pregnancies at a younger age [≤20
years of age; n= 14 of 44 (31.82%)], compared with 10 of 114
(8.8%) for pregnancies at an older age. Although the GN-CHA
did not specify the quantity of alcohol or drugs absorbed during
pregnancy, the percentage of children whose mother drank or
took drugs during her pregnancy (15.2%, n= 24) was inferior to
the prevalence for First Nations’pregnancies in Canada, esti-
mated in a systematic review of six studies
76
: 36.5% for drinking
and 22.1% for binge drinking during pregnancy.
The possible interaction between MeHg and ethanol has been
examined in animal models.
77–80
In one study, ethanol lowered
Hg accumulation in the prefrontal and motor cortex of mice that
were administered chronic low doses of MeHg and ethanol
equivalent to binge drinking; the authors suggest that ethanol
intake could reduce central nervous system Hg levels associated
with psychiatric and cognitive disorders in MeHg-intoxicated
individuals.
80
Future studies in communities with high Hg expo-
sure might consider addressing whether drinking might relieve or
aggravate Hg-related symptoms.
In the present study, an inverse relation was observed between
mothers’age at childbirth and children’s attempted suicide. This is
consistent with the findings of a systematic review and meta-
analysis ofattempted suicide in relation to in utero and perinatal fac-
tors, in which teenage pregnancy was identified as an important risk
factor.
81
The review did not include environmental toxic exposures.
The GN-CHA focused primarily on current health, socioeco-
nomic conditions, and well-being and as a result, no specificinfor-
mation was available on each mother’s (G1) possible adverse
experiences earlier in her life. The descriptions of psychological
suffering in this community following the Hg discharge
1–9
resem-
ble those of other environmental disasters, such as the Deepwater
Horizon spill, where an increase in domestic partner fights, depres-
sion, memory loss, and inability to concentrate was reported for
wives of clean-up workers 2 y after the spill.
82
Palinkas et al.
83
reported that the decline in subsistence activities was an important
risk factor for post-traumatic stress disorder in Alaska Natives
affected by the Exxon Valdez oil spill. Furthermore, children who
lived closer to the Hebei Spirit oil spill in Korea, presented a
higher prevalence of depressive symptoms compared with those
who lived farther away.
84
The authors mention that adult suicides
increased following the spill and suggested that parent psychopa-
thology subsequent to a disaster influences children’s psychologi-
cal health.
84
At the time of the GN-CHA, Statistics Canada reported that
suicide risk among First Nation adults was three times higher
than among the non-Indigenous population.
62
The GN-CHA
use of the same questions as the First Nations Regional Health
Survey 2008/2010,
33
administered to 11,043 First Nation adults
from 216 communities in Canada, allowed us to compare Grassy
Narrows with other First Nation communities. In the First
Nations Regional Health Survey 2008/2010,
33
11.8% of partici-
pants reported that a close friend or family member had commit-
ted suicide in the previous year. In the GN-CHA, 27 (36%) of the
75 mothers (G1) who answered this question responded in the af-
firmative. The prevalence of both suicidal ideation and attempted
suicide during one’s lifetime for mothers in the present study was
also considerably higher compared with those of other First
Nations in Canada who were 18–39 y old at the time of the
survey
33
: Fifty percent (n= 40) among the mothers of Grassy
Narrows, compared with 22% in First Nation communities in
Canada, and 40% (n= 32) vs. 13.1%, respectively.
33
Although prenatal Hg exposure was associated with emotional
and behavioral problems and attempted suicide in today’s children
and youth (G2), this relation was not observed for their mothers
(G1). Several reasons possibly explain this difference. For the
mothers (G1), no individual information was available for other
lifetime factors that could have contributed to suicidal ideation or
attempted suicide, notably the breakdown of the mother’s family in
the years following the discharge; some families would have had
more difficulty coping than others. Many historic descriptions of
Grassy Narrows underline the high prevalence of successful sui-
cides,
1,2,4,5
but no studies were performed on their possible relation
to prenatal or childhood Hg exposure or whether their fathers (G0)
had been fishing guides. A previous study in Grassy Narrows has
linked higher lifetime Hg exposure with premature death (n= 154
of 222).
16
It is possible that women with higher exposures may
have died prior to the GN-CHA.
Although we cannot go back in time, we can certainly ask about
the possible contribution of very high concentrations of prenatal
and childhood Hg to the dramatic increase in youth suicide during
the 10–12 y following the discharge of extremely high levels of
Hg. Erikson,
4
who visited Grassy Narrows in 1979, reported that in
the 11 months between 1977 and 1978, 26 young people between
the ages of 11 and 19 years had attempted suicide.
4
In the same
chapter, he provided a description of Hg poisoning that included
“depression and apathy, memory loss and ... explosive shifts in
mood”, but then went on to explain how alcohol and “the way in
which the Ojibway raise their children”accounted for the striking
disorder in this community. In all the accounts of social disruption
in the Grassy Narrows disaster, no one seriously considered
whether Hg exposure may be a contributing factor, and no one con-
sidered whether drinking might relieve or aggravate Hg-related
symptoms.
In the present study, we examined the possible contribution of
family members having been placed in a residential school, but
these variables were not retained in the SEM pathways. This may
be due to the lack of variance given that almost 80% (n= 59) of
75 mothers reported that at least one of their parents had attended
Environmental Health Perspectives 077001-9 131(7) July 2023
a residential school and they were significantly older than those
whose parents had not been placed in a residential school.
There are important limitations to the present study owing to
limited data availability. Generational SEM is inherently complex
and, in the present study, this was compounded by the small size
of the population, paucity of information on G0 and G1’s early
childhood experiences, and missing data. This is, indeed, a rela-
tively small community; the census data for 2016 indicates that
in Grassy Narrows, there were 160 women between the ages of
20 and 54 y and 175 children 5–19 years of age.
32
No specific in-
formation was available on individual grandfather’s (G0) Hg ex-
posure and behavior in the years following the discharge, and we
relied on published reports and descriptions.
1–6,14
Information on
mothers’childhood (G1) was limited to her childhood fish con-
sumption, her father working as a fishing guide, and schooling
and school performance. Missing Hg exposure data were handled
using estimates, based on larger measurements.
16,20
Children’s
mental health was not assessed, but relied on mothers’percep-
tion. All of these factors contributed to limiting model perform-
ance. To ensure the performance of the SEM, we used a series of
statistical techniques to address different types of error, and
power analyses. We tested the reliability of the pathways, using a
different childhood exposure variable (maternal fish consumption
at 10 years of age) and mother-clustered–based modeling. The
strengths of the associations were verified using logistic regres-
sion models without mediating variables. Although not perfect,
the final SEM provided a plausible portrait of the contribution of
Hg effects over three generations.
Although social disruption certainly played an important role in
the psychological well-being of the people of Grassy Narrows First
Nation following the disaster,
1–6
the findings of the present study
suggest that Hg exposure over three generations likewise contrib-
uted to their mental health. However, to our knowledge, no study
has carefully addressed the possible positive psychological impact
on this community’s fortitude and resilience over the past decades.
In the mid-1970s, Grassy Narrows initiated demonstrations and
then legal proceedings to obtain compensation for the harm that was
done to their community.
1–3
In 1975, community members worked
with M. Harada and his team from Japan to document their neuro-
logical health,
19
and since that time the community has maintained
close relations with the Japanese physicians, who have carried out
several series of examinations.
85
Over the past 50 y, successive
Chiefs and Council and grassroots Grassy Narrows people have
worked with scientists for Hg remediation
86,87
and protection of
their territory.
9
Their actions are marked by memoirs to federal and
provincial parliaments, demonstrations, a hunger strike, and the lon-
gest blockade in Canada to stop logging on their territory.
79
Willow,
writing about the blockade, stated: “The story of the Grassy
Narrows blockade cannot be understood apart from the commun-
ity’s multigenerational struggle to endure in the face of political, cul-
tural, and environmental colonization.”
88
After several years of community lobbying of government
ministries, Grassy Narrows First Nation obtained funding for
their community health assessment, as well as access to their own
historic Hg exposure data. This study was only possible through
the leadership of Grassy Narrows First Nation and collaboration
with an academic research team, based on the principles of own-
ership, control, access, and possession of First Nations’data
[OCAP, a registered trademark of the First Nations Information
Governance Centre (FNIGC)].
89
Several authors have aptly
pointed out that Indigenous leadership throughout the research
process is key to decolonizing health research.
90,91
Understanding the intergenerational harm that was done to
the people of Grassy Narrows should serve to support efforts to
restore the health and well-being that this community enjoyed
prior to the discharge of Hg into the river system of their tradi-
tional territory. From a public health perspective, a community-
based interdisciplinary approach
92,93
would be useful to under-
stand and act upon the social, economic, historical, cultural, phys-
iological, and psychological consequences of this and other
environmental disasters.
Acknowledgments
D.M. was invited by the Grassy Narrows First Nation as
scientific advisor for the GN-CHA. She collaborated on study
design, questionnaire content, data analyses, and preparing reports.
She is principal investigator and main author of the present
research. A.P. created the database, determined the statistical
approaches and performed the analyses. She co-wrote the
manuscript with D.M. M.F. participated in the analyses and
dissemination of the GN-CHA. She participated in the writing and
editing of the manuscript. J.D.S. is the initiator and organizer of the
GN-CHA survey. She shared information on the history and
context across generations and participated in data interpretation.
All authors read and approved the final manuscript.
We thank all of the people of Grassy Narrows First Nation who
organized and carried out the Grassy Narrows Community Health
Assessment (GN-CHA): the community advisory committee, the
fieldwork coordinators, and the surveyors. A special thank you to the
people of Grassy Narrows who participated in the GN-CHA and
agreed to share their biomarker data with us. We salute the resilience
of the Grassy Narrows community who have fought for mercury
justice over the past 50 years. The GN-CHA received financial
support from Health Canada and the Ontario Ministry of Health and
Long-Term Care, and technical support from these ministries and the
Ontario Agency for Health Protection and Promotion, the Ontario
Ministry of Indigenous Relations and Reconciliation, and the
Northwestern Health Unit of Ontario. The present study was funded
by the Canadian Institutes for Health Research (152882).
Ethics approval for the GN-CHA was obtained from the
Manitoulin Anishinaabek Research Review Committee (MARRC),
who issued an ethics certificate on 11 November 2016. The study was
conducted according to the guidelines of the Declaration of Helsinki
and approved by the institutional review board of the Université du
Québec à Montréal (2016_e_1350; 6 September 2016) and Health
Canada Research Ethics Board (REB 2017–0006; 3 August 2017).
Informed consent forms were signed by all participants.
Consent included survey participation and the linking of the
information from the GN-CHA to data obtained from previous
surveillance programs or studies of Hg exposure.
The data sets generated and analyzed in the present study are
the property of the Grassy Narrows First Nation. Permission for
use of the data lies with the Grassy Narrows Chief and Council.
References
1. Usher PJ. 1979. The Economic and Social Impact of Mercury Pollution on the
Whitedog and Grassy Narrows Indian Reserves, Ontario. Ottawa, Ontario,
Canada: P.J. Usher Consulting Services.
2. Shkilnyk AM, Erikson K, Miyamatsu H. 1985. A Poison Stronger than Love: The
Destruction of an Ojibwa Community. New Haven, CT: Yale University Press.
3. Vecsey C. 1987. Grassy Narrows Reserve: mercury pollution, social disruption,
and natural resources: a question of autonomy. Am Indian Q 11(4):287–314,
https://doi.org/10.2307/1184289.
4. Erikson K. 1994. A New Series of Trouble: The Human Experience of Modern
Disasters. New York, NY: W.W. Norton, 27–158.
5. Wheatley B, Wheatley MA. 2000. Methylmercury and the health of indigenous
peoples: a risk management challenge for physical and social sciences and
for public health policy. Sci Total Environ 259(1–3):23–29, PMID: 11032132,
https://doi.org/10.1016/s0048-9697(00)00546-5.
6. Etkin D. 2015. Grassy Narrows. In: Disaster Theory: An Interdisciplinary
Approach to Concepts and Causes. 1st ed. Waltham, MA: Elsevier, Butterworth-
Heinemann, 138–142.
Environmental Health Perspectives 077001-10 131(7) July 2023
7. Mosa A, Duffin J. 2017. The interwoven history of mercury poisoning in Ontario
and Japan. CMAJ 189(5):E213–E215, PMID: 27920011,https://doi.org/10.1503/
cmaj.160943.
8. Willow AJ. 2019. Strategies for survival: First Nations encounters with environ-
mentalism. In: Environmental Activism on the Ground: Small Green and
Indigenous Organizing. Clapperton J, Piper L. eds. Calgary, Alberta, Canada:
University of Calgary Press, 23–45.
9. Willow AJ. 2012. Strong Hearts, Native Lands: The Cultural and Political
Landscape of Anishinaabe Anti-Clearcutting Activism. Albany, NY: State
University of New York Press.
10. Neff MR, Bhavsar SP, Arhonditsis GB, Fletcher R, Jackson DA. 2012. Long-term
changes in fish mercury levels in the historically impacted English–Wabigoon
River system (Canada). J Environ Monit 14(9):2327–2337, PMID: 22785387,
https://doi.org/10.1039/c2em30324h.
11. Kinghorn A, Solomon P, Chan HM. 2007. Temporal and spatial trends of mer-
cury in fish collected in the English–Wabigoon river system in Ontario, Canada.
Sci Total Environ 372(2–3):615–623, PMID: 17161450,https://doi.org/10.1016/j.
scitotenv.2006.10.049.
12. Rudd JWM, Kelly CA, Sellers P, Flett RJ, Townsend BE. 2021. Why the English–
Wabigoon river system is still polluted by mercury 57 years after its contamina-
tion. Facets (Ott) 6:2002–2027, https://doi.org/10.1139/facets-2021-0093.
13. Wheatley B, Paradis S. 1995. Exposure of Canadian aboriginal peoples to
methylmercury. Water Air Soil Pollut 80(1–4):3–11, https://doi.org/10.1007/
BF01189647.
14. Wheatley B, Paradis S, Lassonde M, Giguere MF, Tanguay S. 1997. Exposure
patterns and long term sequelae on adults and children in two Canadian indig-
enous communities exposed to methylmercury. Water Air Soil Pollut 97(1–
2):63–73, https://doi.org/10.1007/BF02409645.
15. Wheatley B, Barbeau A, Clarkson TW, Lapham LW. 1979. Methylmercury poi-
soning in Canadian Indians—the elusive diagnosis. Can J Neurol Sci 6(4):417–
422, PMID: 543984,https://doi.org/10.1017/s0317167100023817.
16. Philibert A, Fillion M, Mergler D. 2020. Mercury exposure and premature mor-
tality in the Grassy Narrows First Nation community: a retrospective longitudi-
nal study. Lancet Planet Health 4(4):e141–e148, PMID: 32353294,https://doi.org/
10.1016/S2542-5196(20)30057-7.
17. Eyl T. 1971. Organic-mercury food poisoning. N Engl J Med 284(13):706–709,
PMID: 4925930,https://doi.org/10.1056/NEJM197104012841306.
18. Ekino S, Susa M, Ninomiya T, Imamura K, Kitamura T. 2007. Minamata disease
revisited: an update on the acute and chronic manifestations of methyl mer-
cury poisoning. J Neurol Sci 262(1–2):131–144, PMID: 17681548,https://doi.org/
10.1016/j.jns.2007.06.036.
19. Harada M, Nakanishi J, Yasoda E, Pinheiro MC, Oikawa T, de Assis Guimarâes
G, et al. 2001. Mercury pollution in the Tapajos River basin, Amazon: mercury
level of head hair and health effects. Environ Int 27(4):285–290, PMID:
11686639,https://doi.org/10.1016/s0160-4120(01)00059-9.
20. Philibert A, Fillion M, Da Silva J, Lena TS, Mergler D. 2022. Past mercury ex-
posure and current symptoms of nervous system dysfunction in adults of a
First Nation community (Canada). Environ Health 21(1):34, PMID: 35292021,
https://doi.org/10.1186/s12940-022-00838-y.
21. McQuaid RJ, Bombay A, McInnis OA, Humeny C, Matheson K, Anisman H. 2017.
Suicide ideation and attempts among First Nations peoples living on-reserve in
Canada: the intergenerational and cumulative effects of Indian Residential
Schools. Can J Psychiatry 62(6):422–430, PMID: 28355491,https://doi.org/10.1177/
0706743717702075.
22. Bombay A, McQuaid RJ, Schwartz F, Thomas A, Anisman H, Matheson K. 2019.
Suicidal thoughts and attempts in First Nations communities: links to parental
Indian residential school attendance across development. J Dev Orig Health
Dis 10(1):123–131, PMID: 29923477,https://doi.org/10.1017/S2040174418000405.
23. Bombay A, Matheson K, Anisman H. 2014. The intergenerational effects of
Indian residential Schools: implications for the concept of historical trauma.
Transcult Psychiatry 51(3):320–338, PMID: 24065606,https://doi.org/10.1177/
1363461513503380.
24. Mosby I, Galloway T. 2017. “Hunger was never absent”: how residential school
diets shaped current patterns of diabetes among Indigenous peoples in Canada.
CMAJ 189(32):E 1043–E1045, PMID: 30967370,https://doi.org/10.1503/cmaj.170448.
25. Anderson C. 2020. Grassy Narrows. In: The Canadian Encyclopedia. Last edited
10 December 2020. https://www.thecanadianencyclopedia.ca/en/article/grassy-
narrows [accessed 20 January 2023].
26. Truth and Reconciliation Commission of Canada. 2015. Honouring the Truth,
Reconciling for the Future. Summary of the Final Report of the Truth and
Reconciliation Commission of Canada. Manitoba, Winnipeg, Canada: National
Centre for Truth and Reconciliation, University of Manitoba. https://web-trc.ca
[accessed 20 January 2023].
27. Wilk P, Maltby A, Cooke M. 2017. Residential schools and the effects on
Indigenous health and well-being in Canada—a scoping review. Public Health
Rev 38:8, PMID: 29450080,https://doi.org/10.1186/s40985-017-0055-6.
28. National Research Council; Institute of Medicine Committee on Depression
Parenting Practices, and the Healthy Development of Children. 2009. Depression
in Parents, Parenting, and Children: Opportunities to Improve Identification,
Treatment, and Prevention. Washington DC: National Academies Press.
29. Stern AH, Smith AE. 2003. An assessment of the cord blood:maternal blood
methylmercury ratio: implications for risk assessment. Environ Health Perspect
111(12):1465–1470, PMID: 12948885,https://doi.org/10.1289/ehp.6187.
30. Jacobson JL, Muckle G, Ayotte P, Dewailly
E, Jacobson SW. 2015. Relation of
prenatal methylmercury exposure from environmental sources to childhood IQ.
Environ Health Perspect 123(8):827–833, PMID: 25757069,https://doi.org/10.
1289/ehp.1408554.
31. Lamoureux-Tremblay V, Muckle G, Maheu F, Jacobson SW, Jacobson JL,
Ayotte P, et al. 2020. Risk factors associated with developing anxiety in Inuit
adolescents from Nunavik. Neurotoxicol Teratol 81:106903, PMID: 32512128,
https://doi.org/10.1016/j.ntt.2020.106903.
32. Statistics Canada. 2019. Census Profile, 2016 Census English River 21, Indian
Reserve [Census subdivision], Ontario and Ontario [Province]. https://www12.
statcan.gc.ca/census-recensement/2016/dp-pd/prof/index.cfm?Lang=E [accessed
8 March 2019].
33. FNIGC (First Nations Information Governance Centre). 2012. First Nations
Regional Health Survey (RHS) 2008/10: National Report on Adults, Youth and
Children Living in First Nations Communities.https://fnigc.ca/wp-content/
uploads/2020/09/5eedd1ce8f5784a69126edda537dccfc_first_nations_regional_
health_survey_rhs_2008-10_-_national_report_adult_2.pdf [accessed 12 June
2022].
34. Andrews G, Slade T. 2001. Interpreting scores on the Kessler Psychological
Distress Scale (K10). Aust N Z J Public Health 25(6):494–497, PMID: 11824981,
https://doi.org/10.1111/j.1467-842x.2001.tb00310.x.
35. Legrand M, Feeley M, Tikhonov C, Schoen D, Li-Muller A. 2010. Methylmercury
blood guidance values for Canada. Can J Public Health 101(1):28–31, PMID:
20364534,https://doi.org/10.1007/BF03405557.
36. Mogensen UB, Grandjean P, Heilmann C, Nielsen F, Weihe P, Budtz-Jørgensen
E. 2015. Structural equation modeling of immunotoxicity associated with expo-
sure to perfluorinated alkylates. Environ Health 14:47, PMID: 26041029,
https://doi.org/10.1186/s12940-015-0032-9.
37. Schumacker RE, Lomax RG. 2004. A Beginner’s Guide to Structural Equation
Modeling. 2nd ed. New York, NY: Psychology Press.
38. Sánchez BN, Budtz-Jørgensen E, Ryan LM, Hu H. 2005. Structural equation models.
J Am Stat Assoc 100(472):1443–1455, https://doi.org/10.1198/016214505000001005.
39. Rosseel Y. 2012. lavaan: an R package for structural equation modeling. J Stat
Softw 48(2):1–36, https://doi.org/10.18637/jss.v048.i02.
40. Shook-Sa BE, Chen DG, Zhou H. 2017. Using structural equation modeling to
assess the links between tobacco smoke exposure, volatile organic com-
pounds, and respiratory function for adolescents aged 6 to 18 in the United
States. Int J Environ Res Public Health 14(10):1112, https://doi.org/10.3390/
ijerph14101112.
41. Hu L, Bentler PM. 1999. Cutoff criteria for fit indexes in covariance structure
analysis: conventional criteria versus new alternatives. Struct Equ Modeling
6(1):1–55, PMID: 28946686,https://doi.org/10.1080/10705519909540118.
42. Ullman JB. 2001. Structural equation modeling. In: Using Multivariate
Statistics. Tabachnick BG, Fidel LS, eds. Boston, MA: Pearson Education.
43. Schreiber JB, Nora A, Stage FK, Barlow EA, King J. 2006. Reporting structural
equation modeling and confirmatory factor analysis results: a review. J Educ
Res 99(6):323–338, https://doi.org/10.3200/JOER.99.6.323-338.
44. Textor J, van der Zander B, Gilthorpe MS, Liskiewicz M, Ellison GT. 2016. Robust
causal inference using directed acyclic graphs: the R package ‘dagitty’.IntJ
Epidemiol 45(6):1887–1894, PMID: 28089956,https://doi.org/10.1093/ije/dyw341.
45. Phelps RW, Clarkson TW, Kershaw TG, Wheatley B. 1980. Interrelationships of
blood and hair mercury concentrations in a North American population
exposed to methylmercury. Arch Environ Health 35(3):161–168, PMID: 7189997,
https://doi.org/10.1080/00039896.1980.10667486.
46. Packull-McCormick S, Ratelle M, Lam C, Napenas J, Bouchard M, Swanson H,
et al. 2022. Hair to blood mercury concentration ratios and a retrospective hair
segmental mercury analysis in the Northwest Territories, Canada. Environ Res
203:111800, PMID: 34364863,https://doi.org/10.1016/j.envres.2021.111800.
47. Satorra A, Saris WE. 1985. Power of the likelihood ratio test in covariance struc-
ture analysis. Psychometrika 50(1):83–90, https://doi.org/10.1007/BF02294150.
48. MacCallum RC, Browne MW, Sugawara HM. 1996. Power analysis and deter-
mination of sample size for covariance structure modeling. Psychol Methods
1(2):130–149, https://doi.org/10.1037/1082-989X.1.2.130.
49. MacCallum RC, Browne MW, Cai L. 2006. Testing differences between nested
covariance structure models: power analysis and null hypotheses. Psychol
Methods 11(1):19–35, https://doi.org/10.1037/1082-989X.11.1.19.
50. Moshagen M, Erdfelder E. 2016. A new strategy for testing structural equation
models. Struct Equ Modeling 23(1):54–60, https://doi.org/10.1080/10705511.2014.
950896.
Environmental Health Perspectives 077001-11 131(7) July