zyxwvutsrqponm

zyxwvutsrqponmlk
zyxwvutsrqponm zy
zyxw
zyxwvutsrqponm
zyxwvutsrqponmlkji
zyxw
zyxw
Acru Psychiutr Scand 1996.- 94: 8-17
Printed in U K ull rixhts reserved
Copi rtqht I0 M u d $yuurd I996
ACTA PSYC ti I ATK ICA
SCANDINAVICA
/SSN OfJO/ 6YO.Y

Suicide attempts among Inuit youth: a

community survey of prevalence and risk
factors

zyxwvut
Kirmayer LJ, Malus M, Boothroyd LJ. Suicide attempts among Inuit
youth: a community survey of prevalence and risk factors.
Acta Psychiatr Scand 1996: 94: 8-17. 0 Munksgaard 1996.

The prevalence of and risk factors for attempted suicide and suicidal
ideation were examined with a survey of 99 Inuit, aged 14-25 years,
residing in a community in Northern Quebec. A total of 34% of survey
L. J. Boothroyd'
',
L. J. Kirmayer', M. Malus',

'Division of Social and Transcultural Psychiatry.

Department of Psychiatry, McGill University.
'Department of Family Medicine. McGill University
and Herzl Family Practice Center, Sir Mortimer
B. Davis - Jewish General Hospital and 'Culture
respondents reported a previous suicide attempt, and 20% had attempted and Mental Health Research Unit, Sir Mortimer
B. Davis - Jewish General Hospital, Montreal,
suicide more than once. A suicide attempt had resulted in injury in about
Canada
I l ' % of those surveyed. The prevalence of suicidal ideation was also very
high: 43") of subjects reported past thoughts of suicide, and 26% had had
suicidal thoughts during the month before the survey. Risk factors for
suicide attempts included male gender, having a friend who had attempted
or committed suicide, a history of being physically abused, a history of
solvent abuse, and having a parent with an alcohol or drug problem.
Protective factors included a family history of having received treatment Key words suicide attempt; risk factor;
for a psychiatric problem, more frequent church attendance, and a high epidemiology
level of academic achievement. While individuals in the community who Laurence J Kirmayer, Institute of Community and
are at high risk for suicide can be targeted for preventive measures, the Family Psychiatry. 4333 C6te Ste Catherine Road,
high prevalence and effect of family problems on likelihood of suicide Montreal, PO H3T 1E4. Canada
attempts indicate the need for family- and community-based approaches. Accepted for publication September 30, 1995

Introduction
Across Canada, First Nations and Inuit" young minimum estimate (4, 6). A 1989 US Indian Health
people commit suicide at a rate 5 to 6 times higher Survey found a lifetime prevalence of attempted
than that of their non-aboriginal peers (1 ). suicide of 23.1% among Alaskan Native adolescents,
Although there is wide variation across communit- compared to a value of 16.1% among their peers
ies, the prevalence of attempted and actual suicide from other Amerindian groups ( 5 ) . A community
among the Inuit has been observed to be particularly survey of the Inuit of Nunavik (Northern Quebec)
high (2-5). Many reasons have been proposed for conducted by Sante Quebec in 1992 found a 14%
this, including socio-economic disadvantage, geo- prevalence of attempted suicide in a sample of 618
graphical isolation, rapid culture change with respondents aged 15 years or older (10). When
attendant acculturation stress, and the oppressive adjusted for age, this rate was 3.5-fold higher than
effects of a long history of internal colonialism (6). the prevalence found in a general population survey
However, few studies among Native peoples in of Quebec. The highest rates were found among
North America have used multivariate statistical young people aged 15-24 years, where the preval-
analysis to examine the relative contribution of ence was 27.6% for males and 25.3% for females.
various factors to the risk of suicide attempts (7-9). The factors associated with attempted suicide
Until recently, studies of the prevalence of included a higher level of education, greater expo-
attempted suicide among the Inuit relied on clinical sure to stressful life events, higher levels of psycholo-
and police records and so offered only a crude or gical distress, and a history of drug use (cocaine,
solvents or cannabis).
" In Canada 'First Nations peoples' and 'Inuit' are the preferred While the Sante Quebec data confirm the high
terms for people who are elsewhere referred to as Amerindians prevalence of attempted suicide among the Inuit of
and Eskimo, respectively. Nunavik, the survey did not determine the severity

8
 zy Suicide attempts among Inuit youth

zyxwv
of suicide attempts, and it examined only a limited the Indian Public Health Service in the United
range of risk factors. The present study was under- States (7). The American study used a self-report
taken in order to better estimate the prevalence and questionnaire administered to students in high
risk factors for suicide attempts among Inuit youth school. Since we wished to include young people
by assessing a range of risk factors identified in who were not attending school, and we could not
previous studies among American Indians, Alaska depend on a uniform level of literacy, we adapted
Natives and Aboriginal Canadian populations (6-8, the self-report measure to a face-to-face interview
1 1, 12). We conducted a community survey of 100 format. Items with culture-specific content were
young people in an Inuit settlement in the Arctic of re-worded to fit the Inuit context, and additional
Quebec. During the 2 years prior to this survey, the items were added to tap risk factors identified in an
study community had suffered from a cluster of exhaustive literature review (6). The questionnaire
actual suicides and suicide attempts by young people was translated into Inuktitut, back-translated to
(13). This cluster occurred despite the ready avail- English by a translator who had not seen the
ability of medical and psychiatric services, sug- original English questionnaire, and checked by
gesting the need for research to identify high-risk bilingual speakers and the principal investigator in
groups who were not coming to medical attention, order to resolve any discrepancies. The final ques-
as well as social factors not addressed by conven- tionnaire consisted of 125 items, including sociode-
tional psychiatric models. The study aimed to mographics, family and household characteristics,

zyxwvutsrqpo
identify factors that contribute to this high incidence history of separations and losses, health problems,
of attempted suicide, and to provide information to personal and parental substance use, worries and
serve as a guide for preventive measures. concerns, experience of physical or sexual abuse,
help-seeking, activities and interests.
Material and methods Self-perceived health was assessed with the ques-
tion ‘What is your state of health?’ with four
Sample possible response categories: ‘not good’, ‘could be

zyxwvutsrq
The study subjects were Inuit adolescents and young better’, ‘good’ and ‘very good’. Poor health was
adults, between the ages of 14 and 25 years at the defined as a response of ‘not good’ or ‘could be
time of interview, who were residing in a village in better’.
the Arctic of Quebec (population approximately An index of crowded housing was constructed by
1000 inhabitants). A list of all persons born between dividing the number of people living in the respond-
1966 and 1978 (n = 259) was obtained from hospital ent’s dwelling by the number of bedrooms.
vaccination records, supplemented by the commun- Suicidal ideation and attempts were assessed with
ity housing list. From these sources a random the following questions: ‘have you ever thought of
sample of 100 subjects was drawn. Research assist- committing suicide?’ and ‘have you ever tried to
ants then contacted the subjects to invite them to commit suicide?’. The concept of what constitutes
participate in the study. However, because many of a suicide ‘attempt’ varies widely among individuals,
those invited were camping and hunting out on the and in many cases suicidal ‘gestures’ of little or no
land during the summer months of the study, this lethality are regarded as attempts. In order to
randomization scheme was successful for only 36 estimate the prevalence of ‘serious’ suicide attempts
subjects; the remaining 64 subjects were recruited in our study population, we asked whether respond-
by sampling through acquaintance, and constitute ents were injured, saw a doctor, or were admitted
a convenience sample. We cannot determine to a hospital as a result of an attempt (14).
whether there was any systematic bias in our sample.
While it is possible that young people who were Atienation. The scale used to measure alienation
better integrated into their families and into tradi- from family and community was adapted from the
tional camp life were under-represented in our 9-item scale of Grossman et al. (7), which had
sample because of the time of year of the survey, it excellent internal consistency with a Cronbach’s
is also possible that the most disaffected individuals alpha of 0.85. One item in this scale, which referred
‘made themselves scarce’ or refused to participate. to alienation from people at school, was not
The mean age and percentage of males in the sample included in our study because we focused on a
did not differ significantly from the community age community rather than a school sample. Our scale
cohort as a whole. included the remaining eight questions, in which
subjects were asked to indicate how much they felt
Measurements
certain groups (including elders, parents, church
leaders and neighbours) cared about them, and how
The survey instrument was adapted from the much their family cared about their feelings, under-
Adolescent Health Survey undertaken in 1988 by stood and paid attention to them, and shared times

9
Kirmayer et al. zyxwvutsrq
zyxwvutsr
zyxwvutsr
of fun with them. Responses were rated on a 5-point
Data analysis
scale from ‘a lot’ (1) to ‘not at all’ (5). A higher
score indicates greater feelings of alienation from Differences in means and proportions were tested
these groups. The Alienation Scale showed good with t-test and Chi-square statistics, respectively.
internal consistency in this sample, with a Fisher’s exact test was used for bivariate 2 x 2
Cronbach’s alpha of 0.80. contingency table analysis in which the expected
number of people in any cell was less than five.

zy
Substance use index. In a series of questions, Because of the exploratory nature of this study, the
respondents were asked about the frequency and lack of previous suicide research on the study
recency of use of alcohol, cannabis, stimulants, population, and the importance of maintaining
solvents and other substances, as well as drinking adequate analytical power, a P-value of 10.10 was
and drug-related problems. An index to approxi- used to indicate a significant difference for statistical
mate the DSM-IV criteria for substance use disorder tests in the bivariate analysis, and 0.10 < P I 0.15
was constructed using the existing questionnaire was considered to indicate a trend (15).
items. To be classified as possibly having a substance Logistic regression analysis was conducted with

zyx
use disorder, the respondent had to indicate that he suicide attempt as the dichotomous dependent vari-
or she had used alcohol or other substances in his able and putative risk factors as independent vari-
or her lifetime, and to report incidents resulting ables. Backward stepwise selection was carried out
from alcohol or drug use from at least two of the on each model with variable addition and removal
following groups of experiences: P-values of 0.05 and 0.10, respectively, in order to
find the most parsimonious reduced models for the
failure to carry out personal duties at school prediction of suicide attempt.
or work (specifically, studies suffering and/or a
teacher being treated unfairly, or the respond-
Results
ent being fired from work);
use of substances in hazardous situations (spe- Of 100 interviews conducted, one was of a 26-year-
cifically, the respondent having an accident or old subject, resulting in 99 usable interviews. The
injury while under the influence of substances, languages used during the interviews were Inuktitut
or drinking and driving); in 69Y0, English in 54% and French in 12%. There
legal problems (specifically, the respondent was no significant effect of language of interview
being sent to jail or a detention centre); on reports of suicidal ideation. However, the use of
social or relationship problems (specifically, the English during the interview was associated with a
respondent having argued with family, lost higher frequency of reporting suicide attempts. Of
friends, separated from boyfriend or girlfriend, those using English in the interview, 43% reported
or shown signs of violence). a suicide attempt compared to 22% of those not
using English (x2=4.21, d.f. = 1 , P=0.04). There
was no effect of season of interview (summer vs.
winter) on the reported frequency of suicidal
Interview procedure
attempts or ideation. The interview took 1-2 h for
A total of 14 lay interviewers and translators were most respondents. Although the questionnaire was
trained and then conducted the interviews singly or well tolerated, about 10% of respondents found it
in pairs under the supervision of the co-investigators moderately upsetting. Suicide attempters did not

zyxw
(a family physician and a psychiatrist). Interviews find the questionnaire more upsetting than non-
took place between June and October 1992, and attempters.
were conducted in the respondent’s home, the nurs-
ing station, at work, or at school. Informed consent
Sociodernographic characteristics
was obtained from all respondents. Interviews were
conducted in Inuktitut and/or in English. French The sample was 48% male, with a mean age of 19.4
translations were occasionally provided in order to years (SD=2.9 years); 73% of subjects were single,
clarify a question. Respondents were encouraged to 26% were married or common-law, and none were
use whichever language was most comfortable for divorced or widowed. One-third of the subjects
them, and often switched languages during the (33%) were students and 32% were working full-
interview in order to clarify a question or response. time or part-time. Twelve per cent described them-
When interviewers identified young people with selves as homemakers and 24% were unemployed
intense suicidal ideation, they encouraged them to (less than half of these were looking for work).
visit the community nursing station and, with their Only 4% were receiving welfare benefits. Of those
verbal permission, informed the community worker. who reported a religious affiliation, the majority

10
(88%) were Anglican, 11Y0 were Catholic and 1%
zy Suicide attempts among Inuit youth

reported cannabis use, 18% at least weekly; and

zyxwvut
were Pentecostal. Although only 1% reported ‘no 37% had used solvents, but only 5% of respondents
religion’, 15% stated that they did not know their reported using solvents at least once a month.
religious affiliation. This may reflect uncertainty or According to the DSM-IV-derived Substance Use
agnosticism, or simply difficulty in understanding Index, 40% of the sample were classified as possibly
the question. As an index of religiosity, regular having a substance use disorder. In addition, 38%
church attendance (at least every Sunday) was of respondents reported that a parent had a sub-
reported by 37% of respondents. stance abuse problem. While 15% of respondents
reported having had a psychological problem during
their lifetime, in response to a broader question,
Family and household characteristics
23% reported a personal, emotional, behavioural or
Most respondents were living with their biological mental problem within the last year, and 29.5% had
parents (67%) and/or full siblings (76%), 16% were a family member who had been treated for psychiat-
adopted and 3% were living in a boarding home. ric illness or emotional problems.
Sixteen per cent of respondents had been separated Levels of exposure to attempted and actual sui-
from their parents before the age of 16 years because cide were extremely high. A total of 27% of the
of hospitalization or problems in the family, 11% respondents had a relative who had attempted or
had parents who had separated or divorced, and committed suicide, and 18% had a relative who had
for 13% of subjects one or both parents had died. attempted or committed suicide in the past year,
For 28% of respondents, a family member had been while 48% had a friend who had attempted suicide
treated for a psychiatric or emotional problem, and 42% had a friend who had committed suicide.
most often a brother for either depression or sub-
stance abuse. The modal number of people per
household was six. Age differences
We divided the sample into two cohorts based on
Prevalence of suicide attempts, ideation, and potential risk age at the time of the survey ( 14-19 years old, n =
factors 58; 20-25 years old, n =4 1). There were no signific-
ant differences between the two cohorts in preval-
Of 91 respondents who answered the questions on
ence of suicidal ideation or attempts, injury
suicide attempts, 34% had made a suicide attempt
associated with a suicide attempt or exposure to
in their lifetime, 20% had made more than one
suicidal friends or relatives. However, the younger
attempt, and 12% had made more than two
attempts. For 32% (10/31) of the attempters, the cohort was more likely to report multiple suicide
attempts (28% vs. 8%; x2=5.35, d.f.= 1, P=0.02),

zyxwvu
suicide attempt was serious enough to have resulted
suicidal thoughts during the last 6 months (36% vs.
in some form of injury. A total of 16% (5/31) of
11%; x2=7.17, d.f.=l, P=0.007), and previous or
attempters saw a doctor as a result, and 13% (4/31)
current use of solvents (49% vs. 20%; xz=8.42,
were admitted to hospital.

zyxwvut
d.f. = 1, P=0.004). There were no other significant
A total of 43% of the respondents reported a
differences between age cohorts with regard to
history of suicidal ideation; 27% had had suicidal
sociodemographic variables, personal or parental
ideation in the last 6 months and 18% in the last 3
months. Suicidal ideation varied widely in intensity. substance use, history of psychological problems,
physical or sexual abuse, or the Alienation scale.
While 26% of respondents reported at least some
degree of suicidal ideation in the last month, 5%
reported intense suicidal ideation, and 2% indicated
Gender differences
that they felt like committing suicide ‘at the first

zyxwvu
opportunity’. As an indication of the link between Male subjects were more likely than females to
ideation and action, of those respondents with report a previous suicide attempt (43% vs. 26%;
suicidal ideation in the last month, 68% (17/25) x2=3.15, d.f.=l, P=O.OS), to have used solvents

zyxw
indicated that they had made a previous or recent in their lifetime (46% vs. 29%; x2=2.97, d.f. = 1 ,
suicide attempt, compared to 21% (14/66) of those P=0.09), and to have a positive score on the
without suicidal ideation in the last month (xz=

zy
Substance Use Index (51% vs. 31%; x2=4.22, d.f.=
17.7, d.f.=l, P<O.OOl). 1, P=0.04). On the other hand, female subjects
The survey also identified the prevalence of other were more likely to report sexual abuse (20% vs.
potential risk factors: 27% of respondents reported 6.5%; x2 = 3.72, d.f. = 1, P = 0.05) and having a
having been physically abused and 14% reported relative who had attempted or committed suicide
sexual abuse; 71% reported some alcohol consump- (35% vs. 17%; x2=3.71, d.f.=l, P=0.05). Female
tion but only 6% consumed alcohol weekly; 52% subjects were also more likely to attend church

11
Kirmayer et al. zyxwvutsrq
zyxwvutsrqpo
zyxwvuts
zyxwvut
regularly (48% vs. 25%; x2= 5.36, d.f. = 1, P=0.02).

zy
health problems during the previous year ( P 1 0 . 0 1 ) .
There were no significant gender differences for Suicide attempters were also more likely to have
other variables. used solvents at least every couple of months, and
were less likely to attend church regularly (P10.05).
Attempters had significantly higher scores than non-
Risk and protective factors for attempted suicide
attempters on the Alienation Scale (P< 0.05). There
In Tables 1 and 2 the frequencies of potential risk were trends for suicide attempters to be male, not
factors for respondents who reported suicide attending school, the first-born child, living in more
attempts are compared with the values for those crowded households, physically abused, and to have
who had not attempted suicide. Suicide attempters used mouth wash or cough syrup to ‘get high’
were significantly more likely to report parents with (0.05 <P < 0.15). Essentially the same pattern of
a drinking or drug problem, having friends who risk factors was found for suicidal ideation, except
had attempted or committed suicide, use of solvents that there was no association with male gender.
in their lifetime, and having had personal or mental A number of risk factors for attempted suicide

Table 1 Comparison of suicide attempters and non-attempters

Attempted
(n=31) zyxwvuts
Never attempted
(n=60) Result of significance test

zyxwvutsrqpon
Sociodemographic characteristics
Mean age f S D (years) 18.9*2 9 197k2.9 NS”
Gender 1% male) 61 42 x2=3.15, d.f.=l, P=O.O8
Not attending school (%) 74 58 x2=2.22. d.f =1. P=O 14
Working, full-time or part-time (YO) 39 30 NS
Church attendance (at least every Sunday) (YO) 23 43 xz=3.81. d f . = l , P=O.O5
Mean crowding index k S D 1.77 kO.47 1.58f0.48 t=1.85, df.=89, P=0.07
Family history

zyxwvutsrqpon
First born (%I 40 20 xz=30. d f = l . P=008
Adopted (%) 20 17 NS
Brought up by same people all his/her life (%) 84 88 NS
Parents separated or divorced 1%) 13 8 NS
One or both parents dead (Yo) 10 15 NS
Family history of treatment for a psychiatric problem (YO) 27 31 NS
Parent with a drinking or drug problem I%) 61 26 xz=947, d f =1, P=OOO2
Exposure to suicide
Relative has attempted or committed suicide (%) 32 25 NS
Friend has attempted suicide (YO) 65 37 xZ= 6.05, d.f. = 1, P=0 01
Friend has committed suicide (YO) 62 32 x2=7 12. d.f = 1 , P=O.O08

a NS indicates a result with a f-value of > O 15

Table 2 Comparison of suicide attempters and non-attempters

Attempted Never attempted

(n=31) (n=60) Result of significance test

Substance use
Solvents used at some time in the past (%) 58 23 xz=10.6. d.f.=l. P=OOO1
Solvents used at least every couple of months (%) 13 2.0 Fisher‘s exact test; P=0.04
Cough syrup or mouth wash used to ’get high (%) 19 5.1 Fisher‘s exact test: P=O.O6
Alcohol consumed at least once a week (%) 32 8.3 NS”
Substance Use Index ( % positive) 45 35 NS
Personal characteristics
Poor health (%) 32 20 NS
Personal or mental health problem in the past year (%) 40 12 x2=10.9, d . t . = l . P=O.OOl
Physical abuse 1%) 36 20 x2=2.45. d.t.=l, P=0.12
Sexual abuse (YO) 13 14 NS
Grade point average a8 (%) 50 96 Fisher‘s exact test;
(5/10) (27/28) P=O.O03
Mean Alienation Scale score +SO 236k6.6 20.1 k 6 . 9 t= 2.07, d t. = 77, P= 0.04

a NS indicates a result with a Pvalue of > O 15.

12
 zy Suicide attempts among Inuit youth

found in other studies did not prove to be significant

at the bivariate level of analysis in our sample,
including age, unemployment, having been adopted,
having parents who were separated or dead, having
a relative who had had treatment for a psychiatric
problem or who had attempted or committed sui-
cide, weekly alcohol use, indication of a substance
use disorder, not being brought up by the same
people all one's life, self-perception of poor health,
zyxwv four times more likely to have made a suicide
attempt, and having a friend who had attempted
suicide was associated with a 3.4-fold greater odds
of suicide attempt. Although not quite significant
at the P=0.05 level, greater feelings of alienation
were associated with a greater odds of suicide
attempt in the reduced model.
For Model B, the independent variables consid-
ered were selected from a model of the factors
and having experienced sexual abuse. contributing to suicide among aboriginal peoples,
We used logistic regression analysis to determine based on an extensive literature review (6) and the

zyxwvuts
the independent contribution of risk factors to the work of Thorslund (16, 17) on Inuit in Greenland.
likelihood of a previous suicide attempt, controlling Many of these factors affect whole communities,
for the effects of sociodemographic variables (age and others were not assessed in our questionnaire,
and gender) and other independent variables in the and so could not be studied using our data. Nine
model. Three models were examined with logistic variables in our survey which approximated those
regression, using backward stepwise elimination of in the suicide model were entered in the full regres-
non-significant variables (Table 3 ) . sion model, in addition to age and gender. These
Model A included age and nine additional vari- variables were physical abuse, sexual abuse, a family
ables which were found to be significant risk factors history of treatment for a psychiatric problem, a
for suicide attempts among Navajo high-school parent with a drinking or drug problem, a relative
students in a logistic regression analysis conducted who had attempted or committed suicide, a friend
by Grossman et al. (7). These variables were gender, who had attempted suicide, a friend who had com-
sexual abuse, physical abuse, the Alienation Scale, mitted suicide, the Alienation Scale, and the
a friend having attempted suicide, a relative having Substance Use Index.
attempted or committed suicide, perception of poor In reduced Model B, having a friend who had

zyxwvu
health, regular alcohol use (at least once a week), attempted suicide and male gender were highly
and a history of a psychological problem. significant risk factors associated with a 6- to 8-fold
In the reduced version of Model A, male gender greater odds of suicide attempt, whereas older age
and having a friend who had attempted suicide was a significant protective factor. The experience
were both significantly associated with an increased of physical abuse was also marginally significant as
likelihood of suicide attempt, when adjusted for the a risk factor (P=0.053). A family history of treat-
effects of the other variables. Males were nearly ment for a psychiatric problem was a significant

Table 3 Logistic regression of risk factors for suicide attempts among young Inuit

Variables in reduced model Bivariate Logistic

analysis regression Significance
Crude Adjusted level in reduced
odds ratio odds ratioa logistic

zyxwvu
(95% confidence (95% confidence regression
intewal) interval) model
~

Model A n=77
Male gender 2 2 (0.91, 5.4) 3.7 (1 2, 11.4) P=O 02
Friend has attempted suicide 3.1 (1.2. 7.6) 3 4 (1.1, 10.4) P=003
Alienation Scale score 1 1 (0 99. 1.2) P=O 055
Model 8 n=70
Age 0 75 (0.58, 0 96) P= 0 022
Male gender 2.2 (0.91. 5.4) 8.5 (1.9. 38.0) P= 0 005
Friend has attempted suicide 3.1 (1.2, 7.6) 6.1 (1.5. 25 1) P=O 012
Family history of treatment for a psychiatric problem 0.79 (0.3. 2 1) 0.14 (0.03, 0.74) P= 0 021
Physically abused 2.2 (0.82. 5.7) 4.2 (0.98, 18.2) P= 0 053
Model Cb
Solvents used at any time

Physically abused zyxwvutsr

zyxwvutsrqpon
Parent(s) with a drinking or drug problem

"Adjusted for the other independent variables in the final model

Controlling for age and gender
4 6 (1.8. 11.8)
4 3 (1.7. 11.3)
2.2 (082. 5 7)
n=80
7 6 (2 0.28.6)
5.9 (1 8, 19 7)
4.9 (1.2. 19.8)
P= 0 003
P=O 004
P= 0 024

13
 zyxwvutsrqp
Kirmayer et al.

protective factor, perhaps indicating that recogni-

tion and treatment of the problem was beneficial
for the respondent.
Prevalence of suicide attempts and ideation
One-third (34%) of the Inuit young people in our

zyxwvut
Model C consisted of the six variables that sample had made a suicide attempt at some time in

zyxwvut
remained significant in the reduced versions of their life. As an index of seriousness, 11% of subjects
Models A and B, as well as two variables (use of reported that a suicide attempt had resulted in
solvents and a parent with a drinking or drug injury, 5% had seen a doctor as a result of an
problem) that were significant at the bivariate level attempt, and 4% were admitted to hospital. Thus,
after a Bonferroni correction for multiple compar- for one-third of suicide attempters at least one
isons was applied (i.e. P S O . 10/27 = 0.004). This attempt could be considered to be serious.
model allowed us to examine the best predictors of The prevalence of suicide attempts in this sample
suicide attempt from among eight variables, namely was much higher than the values reported in similar
age, gender, a friend having attempted suicide, the age and gender cohorts of the general population
Alienation Scale, a family history of treatment for in North America. For comparison, the US
a psychiatric problem, physical abuse, having used Adolescent Health Survey of high-school students

zyxwvutsrqp
solvents at any time, and a parent with a drinking found a lifetime prevalence of attempted suicide of
or drug problem. Because of missing data in the 14% ( 18). Among 13 000 American Indian and
full model, the number of subjects decreased to 67. Alaskan Native high-school students living on reser-
In order to estimate the adjusted odds ratios for vations or in rural communities, the prevalence was
these factors on a larger proportion of our study 17% ( 5 ) . A similar survey of adolescent Amer-
sample (n = SO), the three significant variables in indians in Alberta found a prevalence of 25% among
the reduced version of Model C were entered into those aged 15 years or older (9). Still higher rates
a model with age and gender as control variables. have been reported among special populations,
Of the significant risk and protective factors

zyxwvu
including Amerindian adolescents in residential
identified in Model C, the strongest predictor of schools. For example, a survey of 83 freshman
suicide attempt was the use of solvents at any time, students from the Zuni Public High School in New
which was associated with a 7.6-fold greater odds Mexico found a lifetime prevalence of suicide
of suicide attempt, followed by a parent with a attempts of 30% (19).
drinking or drug problem (odds ratio=5.9), and
Meehan et al. (14) used the same questions as
experience of physical abuse (odds ratio =4.9).

zyxwv
those employed in the present study to assess the
seriousness of suicide attempts among 18- to
Discussion 24-year-old university freshmen in the US. They
This study is the first to examine risk factors for found a self-reported lifetime prevalence of
attempted suicide among Inuit young people in the attempted suicide of 1OYn. As indices of seriousness,
community using epidemiological survey data and 4.6% of subjects reported having been injured in an
multivariate statistical analysis. Although we were attempt, 3% had sought medical attention as a
not able to obtain a strictly random sample, the result of an attempt, and 1% were hospitalized
age and gender characteristics of our sample did because of a suicide attempt. Compared to the
not differ significantly from the age cohort in the suicide attempters among US college freshmen,
community as a whole. Unlike the US study of then, a smaller proportion of young Inuit who made
Navajo adolescents by Grossman et al. (7), we were suicide attempts were injured or saw a doctor as a
able to include young people who were not result, but they were equally likely to be hospital-
attending school; only one-third of our sample were ized. This may indicate a higher proportion of less
students. The study sample represented almost 40% serious attempts among the Inuit, combined with
of the people between the ages of 14 and 25 years easier access to hospital care through the nursing
in the community. While it is possible that some of station.
the most troubled young people avoided the survey Suicidal ideation was also very common among
or refused to participate, our sample is representat- young Inuit, with 43% reporting a past history, 26%
ive of a broad range of youth. Respondents who reporting some suicidal ideation during the last
used English in the interview were more likely to month and 5% reporting intense suicidal ideation.
report suicide attempts than those who did not use Consistent with the notion of a spectrum of suicidal
English. This may represent a greater willingness to behaviour, suicidal ideation was significantly associ-
disclose suicide attempts to outsiders, greater emo- ated with suicide attempts ( 5 , 20, 21).
tional distance involved in using a second language, The prevalence of suicide attempts and ideation
or a substantive effect of linguistic acculturation on in the present study is substantially higher than the
increased suicide risk. values found in the 1992 Sante Quebec Inuit survey

14
 24-year-old people had a lifetime prevalence of
26.5% and a period prevalence of 17.9% in the
zy
(10). In that survey, suicide attempts among 15- to
Suicide attempts among Inuit youth

roles experienced by Inuit men, resulting in greater

acculturation stress (23, 24). Female subjects were
more likely to have been sexually abused, but this
previous month. Suicidal ideation had a lifetime factor was not significantly related to suicide

zyxwvuts
zyxwvu
prevalence of 12% among the Inuit as a whole and attempts.
16.2% in the 15- to 24-year-old age cohort. These
rates are actually lower than the rates of attempted
Exposure to suicide
suicide in the same survey. This counter-intuitive
finding probably reflects the question used in the Not surprisingly, given the high prevalence of sui-
survey, which asked ‘Have you ever seriously cidal behaviour in the community, 57% of young
thought of committing suicide (killing yourself )?’ people had a friend who had attempted or commit-
Apparently respondents interpreted the qualifier ted suicide. A total of 25% had relatives who had
‘seriously’ as equivalent to having actually attempted or committed suicide, and 18% had relat-
attempted suicide. Accordingly, the present estim- ives who had attempted or committed suicide in the
ates of suicidal ideation appear to be more meaning- past year. These very high levels of exposure to
ful than those of the Sante Quebec survey. suicidal behaviour probably contribute directly to
It should be noted, however, that the Sante the tendency for suicide to occur in clusters, as a
Quebec study surveyed the territory of Nunavik as type of chain reaction, in a community where so
a whole (sampling 400 households in 14 settle- many people are closely related (25). This points
ments), while the present survey examined only one to the need to consider suicide not just as an
community which had suffered a cluster of suicides individual problem but as a dilemma at the level of
in recent years. Our results can therefore be consid- the community (26). This can be addressed in future
ered to be indicative of the impact of a suicide research by comparative studies of communities
cluster on young people in a comparatively large which vary widely in the frequency of actual suicide
Inuit settlement at one particular moment in time. (1, 4).

Age effects Risk factors for attempted suicide

When we divided the sample into two age cohorts Comparison of young people who had made suicide
(14-19 and 20-25 years of age), it was found that attempts with those who had never attempted sui-
the younger group was more likely to report mul- cide revealed the following risk factors at the bivari-
tiple suicide attempts, suicidal thoughts during the ate level: male, first-born child, no longer at school
previous 6 months, and use of solvents in the past. or doing relatively poorly at school, attending
The increased likelihood associated with younger church less frequently, living in a relatively crowded
age was supported by the logistic regression analyses house, having a parent with a drinking or drug
discussed below. This finding may indicate a greater problem, having experienced physical abuse, having
vulnerability of younger adolescents to the social used solvents at some time in the past, and having
factors which affect the whole community. used cough syrup or mouth wash to ‘get high’.
Suicide attempters were more likely to report a
personal or mental health problem in the previous
Gender effects
year and to feel alienated from the community and
There were few significant gender differences. In the their families. Attempters were about twice as likely
general population, females are more likely to make as non-attempters to have a friend who had
suicide attempts, while males are more likely to attempted or committed suicide, although there was
commit suicide; this difference is usually attributed no such difference in exposure to relatives who had
to a tendency for male subjects to use more lethal attempted or committed suicide.
methods when attempting suicide (22). In this Several of these variables may also be interpreted
sample of young Inuit, males were actually more as protective factors. For those subjects still at
likely to make suicide attempts. The proportion of school, having an average grade of B or greater was
attempts that resulted in injury was not significantly associated with a lower risk of suicide attempt.
different between males and females. Male subjects Success at school is a source of self-esteem as well
were more likely to have used solvents in the past, as an indicator of general problem-solving ability
and the relationship between use of solvents and and coping skills. Regular church attendance also
suicide attempts may have accounted for much of appeared to have a protective effect against the risk
their excess risk. A more remote cause, not assessed of suicide attempts. Quality of family life and
in the present study, may be the greater discontinuit- religiosity are highly correlated in the general popu-
ies between traditional and contemporary social lation (27). The impact of religion on suicide

15
Kirmayer et al. zyxwvutsrq
attempts may be understood not so much in terms
of specific beliefs about suicide, suffering and the
afterlife, but more in terms of the way in which
by 6 times. In the final reduced model (C in Table 3),
the significant risk factors were solvent use, a parent
with a drinking or drug problem and physical abuse.
religious affiliations and practices organize social These results point to the role of family problems
support networks (28). Religiosity may reduce the in setting the stage for suicide attempts among
suicide rate by strengthening social ties through young people ( 5 ) .
participation in community activities. This is con- Recently, some authors have related the high
sistent with the findings of the present study, in prevalence of suicide among contemporary Inuit
which feelings of alienation and parental substance youth to a traditional pattern of ‘altruistic’ suicide
abuse were found to be risk factors for suicide (29, 30). They argue that, in the face of rapid
attempts. culture change, disorientation and loss, young Inuit

zyxwvutsr
A number of risk factors identified in other deal with feelings of hopelessness by following a
studies were not found to be significant at the path of self-destruction that, through its evocation
bivariate level in this sample, including unemploy- of traditional altruistic suicide, affirms their Inuit
ment, family psychiatric history, having a relative identity (10). Such cultural hypotheses are difficult
who had attempted or committed suicide, alcohol to test. However, the majority of young and adult
consumption and sexual abuse ( 5 ) . These negative Inuit interviewed in parallel ethnographic research
findings should be interpreted with caution, since by our team viewed suicide as aberrant, and did
in some cases respondents may not have answered not associate it with Inuit traditions of self-sacrifice
questions truthfully (especially questions about for family or group survival (31). It seems likely
sexual abuse and family psychiatric history). then that identity confusion, dislocation, power-
Furthermore, certain other potentially important lessness and loss engendered by rapid culture change
factors were not included in the questionnaire, and internal colonialism (32-36) are the more signi-
notably questions directed at specific psychiatric ficant social determinants of the hopelessness and
morbidity and at social problems such as family despair that lead to suicide attempts among Inuit
conflict or violence (although exposure to physical young people.
abuse was measured and found to be a risk factor The prevalence of suicide attempts among young
in subsequent multivariate analyses). Inuit in a Northern Quebec community is among
The use of logistic regression allowed us to estini- the highest reported in the literature: one-third of
ate the effects of specific risk factors while con- the young people surveyed had attempted suicide,
trolling for possible confounding variables. and almost half had a friend who had attempted
Unfortunately, the size of our sample only permitted suicide. This community-wide prevalence points to
us to work with about 10 variables. We selected the need to consider both individual and community

zy
potential risk factors to enter in logistic regression factors in accounting for the elevated rate. In this
models not only from those that were statistically study, we could only address individual factors with
significant at the bivariate level, but also from those the data, since we only surveyed one community.
that had been found to be important in previous Further consideration of community factors would
research, that were important in theoretical models, require the study of multiple communities in the
and that had the greatest potential clinical and same time frame in order to identify the social factors
public health importance. We tested a series of that contribute to differentially elevated rates.
models, each of which provides useful information.
These models indicated that males were about 4 Acknowledgements
to 8 times more likely than females to have made a
suicide attempt; younger respondents were 1.3 times This research was supported by grants from the Kativik
Regional Board of Health and Social Services (‘Suicide risk
more likely to have made a suicide attempt. Young among Inuit youth’), the Conseil Quebecois de la Recherche
people with a friend who had attempted suicide Sociale (‘Inuit concepts of mental health and illness’) and the
were 3 to 6 times more likely to have made an Fonds de la Recherche en Sante du Quebec (‘Culture and mental
attempt themselves, and subjects exposed to phys- health’). We also thank the staff and administration of
ical abuse were 4 to 5 times more likely to make a Inuulitsivik Health Center and the Institute of Community and
Family Psychiatry for their support throughout this research.
suicide attempt than individuals who had not been The project was made possible by the close collaboration of
subjected to such abuse. A family history of treat- Harry Tulugak, Noah-Adamie Qumaaluk and Dr Andre
ment for a psychiatric problem reduced the odds of Corriveau. We thank our research assistants in the field, Melina
suicide attempt by a factor of almost 1/10. A history Auerbach, Clara Aupaluk, Martha Aupalu, Christopher
of solvent abuse was associated with an increase of Fletcher, Shao Chyi Lee, Lisa Nashuak, Trina Qumaluk, Robert
Watt and Kate Zimmerman, who ably assisted with recruiting
about 8-fold in the odds of attempted suicide, while subjects and administering the interview. Consuelo Quesney,
a history of drug or alcohol abuse by the respond- Suzanne Taillefer and Terri Yu assisted with the coding, entry
ent’s parents increased the odds of suicide attempt and analysis of data.

16
References
zy
zyxwvutsrqponm
I . Royal Commission on Aboriginal Peoples. Choosing life:
special report on suicide among aboriginal people.
19. HOWARD-PITNEY
Suicide attempts among Inuit youth
B, LAFROMBOISE TD, BASILM et al.
Psychological and social indicators of suicide ideation and
suicide attempts in Zuni adolescents. J Consult Clin Psychol
1992: 60: 473-476.
Ottawa, 1995. 20. PEARCE CM, MARTING. Predicting suicide attempts among
2. KETTL PA, BIXLER EO. Suicide in Alaska natives, adolescents. Acta Psychiatr Scand 1994: 90: 124-128.
1979- 1984. Psychiatry 1991: 54: 55-63. 21. LEWINSOHN PM, ROHDEP, SEELEYJR. Psychosocial risk
3. RODCERS DD. Suicide in the Canadian Northwest factors for future adolescent suicide attempts. J Consult
Territories 1970-1980. In: HARVALD R, HART-HANSEN JP, Clin Psychol 1994: 62: 297-305.
ed. Circumpolar Health 81. Nordic Council for Arctic 22. GARRISONCZ. Demographic predictors of suicide. In:
Medical Research Report, 1982: 492-495. MARISRW et al., ed. Assessment and prediction of suicide.
4. PETAWABANO B, GOURDEAU E, JOURDAIN F et al. Mental New York: Guilford, 1992: 484-498.
health and aboriginal people of Quebec. Montreal: Gaetan 23. MATTHIASSON Js. Living on the land: change among the
Morin Editeur, 1994. Inuit of Baffin Island. Peterborough, Ontario: Broadview
BLUMRW, HARMON B, HARRISL et al. American Indian- Press, 1992.
Alaska Native youth health. JAMA 1992: 267: 1637-1644. 24. MCELROYA. Canadian Arctic modernization and change
KIRMAYER LJ. Suicide among Canadian Aboriginal peoples. in female Inuit role identification. Am Ethnol 1975: 24:
Transcult Psychiatr Res Rev 1994: 31: 3-58. 662-686.
GROSSMAN DC, MILLICAN BC, DEYORA. Risk factors for 25. BECHTOLD DW. Cluster suicides in American Indian adoles-

zyxwvutsrqpon
suicide attempts among Navajo adolescents. Am J Public cents. Am Indian Alaska Native Ment Health 1988: 1:
Health 1991: 81: 870-874. 26-35.
8. MANSONSM, BEALSJ, DICK RW et al. Risk factors for 26. RODGERSDD. Community crisis intervention in suicide
suicide among Indian adolescents at a boarding school. epidemics. In: POSTLB et al., ed. Circumpolar Health 90.
Public Health Rep 1989: 104: 609-614. Winnipeg: University of Manitoba Press, 1991: 276-280.
9. GARTRELL JW, JARVIS GK, DERKSEN L. Suicidality among 27. STACKS. Marriage, family, religion, and suicide. In: MARIS
adolescent Alberta Indians. Suicide Life Threat Behav 1993: RW et al., ed. Assessment and prediction of suicide. New
23: 366-373. York: Guilford, 1992: 540-552.
10. BUYER R, DUFOURR, PRBVILLE M et al. State of mental 28. PESCOSOLIDO B, GEORGIANNA S. Durkheim, suicide and
health. In: QUEBECS, JETTEM, ed. A health profile of the religion: toward a network theory of suicide. Am Sociol
Inuit: report of the SantC Quebec Health Survey among the Rev 1989: 54: 33-48.
Inuit of Nunavik, 1992. Montreal: Ministere de la Sante et 29. BALIKCIA. In: BLISHENB et al., ed. Suicidal behavior
des Services Sociaux, Gouvernement du Quebec, 1994: among the Netsilik Eskimos, in Canadian Society: sociolo-
117-144. gical perspectives. Chicago: Free Press of Glencoe, 1961.
11. RYLAND DH, KRUESIMJP. Suicide among adolescents. Int 30. LEIGHTONA, HUGHESCC. Notes on Eskimo patterns of
Rev Psychiatry 1992: 4: 185-195. suicide. Southwest J Anthropol 1955: 11: 327-338.
12. EARLSF, ESCOBARJI, MANSONSM. Suicide in minority 31. KIRMAYER LJ, FLETCHER C, CORINE et al. Inuit concepts
groups: epidemiologic and cultural perspectives. In: of mental health and illness (working paper no. 4). Sir
BLUMENTIIAL SJ et al., ed. Suicide over the life cycle. Mortimer B. Davis Jewish General Hospital: Culture and
-

Washington, DC: American Psychiatric Press, 1991: Mental Health Research Unit, Department of Psychiatry,
571 -598. 1994.
13. KIRMAYER LJ, GRIN E, CORRIVEAU A et al. Culture et 32. O N E I L JD. Colonial stress in the Canadian Arctic: an
maladie mentale chez les Inuit du QuCbec. Sante Mentale ethnography of young adults changing. In: JANESCR et al.,
au Quebec 1993: 18: 53-70. ed. Anthropology and epidemiology. Dordrecht: D. Reidel,
14. MEEHANPJ, LAMBJA, SALTZMAN LE et al. Attempted 1986: 249-274.
suicide among young adults: progress toward a meaningful 33. CONDONRG. Adolescence and changing family relations in
estimate of prevalence. Am J Psychiatry 1992: 149: 41-44. the Central Canadian Arctic. Arctic Med Res 1990: 49:
15. GROVEWM, ANDREASSEN N. Simultaneous tests of many 81-92.
hypotheses in exploratory research. J Nerv Ment Dis 1982: 34. VALLEEF. Stresses of change and mental health among the
170: 3-8. Canadian Eskimos. In: NAGLERM, ed. Perspectives on the
16. THORSLUND J. Inuit suicides in Greenland. Arctic Med Res North American Indians. Toronto: McClelland & Stewart,
1990: 49: 25-34. 1972: 203-213.
17. THORSLUND J. Suicide among Inuit youth in Greenland 35. DUFFYRQ. The road to Nunavut: the progress of the
1974-86. In: POSTL et al., ed. Circumpolar Health 90. Eastern Arctic Inuit since the Second World War. Kingston:
Winnpeg: University of Manitoba Press, 1991: 299-302. McGill-Queen’s Press, 1988.
18. American School Health Association. The Nationat 36. BERRYJ W. Acculturation among circumpolar peoples:
Adolescent Health Survey: a report on the health of implications for health status. Arctic Med Res 1985: 40:
America’s youth. Oakland, CA: Third Party Publishing, 21-27.
1989.

17