Suicide Among Inuit: Results From a Large, Epidemiologically Representative Follow-Back Study in Nunavut

Abstract

Objective:

The Inuit population in Canada’s North has suffered from high rates of death by suicide. We report on the first large-scale, controlled, epidemiologically representative study of deaths by suicide in an Indigenous population, which investigates risk factors for suicide among all Inuit across Nunavut who died by suicide during a 4-year period.

Methods:

We identified all suicides by Inuit (n = 120) that occurred between January 1, 2003, and December 31, 2006, in Nunavut. For each subject, we selected a community-matched control subject. We used proxy-based procedures and conducted structured interviews with informants to obtain life histories, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Axis I and II diagnoses, and measures of impulsive and (or) aggressive traits.

Results:

Compared with control subjects, subjects who died by suicide were more likely to have experienced childhood abuse (OR 2.38; 95% CI 1.39 to 4.08), have family histories of major depressive disorder (P = 0.002) and suicide completion (P = 0.02), and have been affected by major depressive disorder (OR 13.00; 95% CI 6.20 to 27.25), alcohol dependence (OR 2.90; 95% CI 1.59 to 5.24), or cannabis dependence (OR 3.96; 95% CI 2.29 to 6.8) in the last 6 months. In addition, subjects who died by suicide were more likely to have been affected with cluster B personality disorders (OR 10.18; 95% CI 3.34 to 30.80) and had higher scores of impulsive and aggressive traits (P < 0.001).

Conclusions:

At the individual level, clinical risk factors for suicide among Inuit are similar to those observed in studies with the general population, and indicate a need for improved access to mental health services. The high rate of mental health problems among control subjects suggests the need for population-level mental health promotion.

The World Health Organization estimates that suicide accounts for almost 1 million deaths per year, worldwide,¹ and for men between the ages of 15 and 44 years, suicide ranks among the leading causes of death.² However, nowhere does suicide have such an impact and social burden as among indigenous populations. Although many indigenous communities have suicide rates that are comparable to those of their country’s general populations, many groups have much higher suicide rates.³ In Canada, the rates of suicide among indigenous people have been consistently higher than in the general population.⁴

Inuit suicide rates are among the highest in the world. Between 1999 and 2003, the rates in Inuit regions averaged 135 per 100 000, more than 10 times higher than the general Canadian rates.⁵ Although suicide was not unknown in Inuit culture, evidence suggests this rate has increased severalfold during the last decades. Moreover, this rise is almost entirely due to an exponential increase in suicides by people younger than 25 years old.⁶

About 60 000 people in Canada identify themselves as Inuit, and nearly one-half live in Nunavut.⁷ Nunavut is the largest and northernmost territory of Canada and was created via the Land Claims Agreement, which established a public territorial government with the aim of ensuring self-reliance and the protection of Inuit values and traditions. It comprises 26 remote communities, situated north of the 60th parallel (the capital Iqaluit is located at about lat 63°75′N and long 68°55′W), and is only accessible by plane or boat in the summer.

The extremely high rates of suicide in many indigenous communities contrast with the paucity of empirically derived data available. There are some studies of suicide attempts, but very few of suicide completions among indigenous people, a significant lack considering that these 2 forms of suicidal behaviour differ in important ways.⁸ This discrepancy may be partially explained by the great logistical challenges of conducting studies in isolated communities.

Clinical Implications

• Previously identified contributors to suicide and suicidal behaviour, including early-life adversity, are associated with suicide in Nunavut.

• Distal-acting strategies, such as interventions targeting parenting styles in young families, could improve long-term mental outcomes of the Inuit population.

• Proximal-acting strategies, such as developing mental health awareness and treatment programs for youth, may play an important role in suicide prevention.

Limitations

• Our study examined individual suicide risk factors, rather than community- or population-level factors linked to significant social changes during recent decades.

• We specifically investigated Inuit in Nunavut; it is unclear how well these observations can be generalized to other Indigenous communities or to other regions.

Although a consistent body of evidence supports the strong association between psychopathology and suicide,⁹ the role of psychopathology has been questioned, particularly in some non-Western contexts.^10,11 Social structural factors may contribute to elevated levels of suicide in a whole population. Social explanations for suicide among indigenous peoples in Canada stress the effects of colonization, sedentarization, relocation, and forced assimilation through residential schools.¹² Of course, social and psychopathological explanations are not mutually exclusive; both may play a role and interact to increase the risk of suicide. Indeed, they may be directly linked in that social adversity may increase the risk of mental health problems associated with increased risk for suicide. Nevertheless, the prevalence of psychopathology among people who die by suicide has not been systematically examined in Indigenous communities with high rates of suicide, including the Inuit.

Here we report on the first large-scale, controlled, epidemiologically representative and proxy-based study of deaths by suicide conducted with an indigenous population. The Qaujivallianiq Inuusirijauvalauqtunik [Learning From Lives That Have Been Lived] was an age- and sex-matched case–control study aiming to understand risk factors for suicide in all Inuit people who died by suicide between 2003 and 2006 across Nunavut.

Methods

Our study was approved by the Institutional Review Board of the Douglas Mental Health University Institute, and the Nunavut Research Institute issued a research license. The project was developed in partnership with Nunavut community organizations.¹³

A lay report was developed and presented to the stakeholders and community members for knowledge dissemination and approval by the stakeholders before the submission for publication, as required by chapter 9 of the second edition of the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.¹⁴

Study Population

Nunavut’s population in 2011 was 31 906, with a median age of 24.1 years (compared with 40.6 years in the whole of Canada).¹⁵ The vast majority of Nunavut’s population is Inuit (83.92%), and 67.8% speak Inuktitut as their mother tongue.^7,15 Our study included all suicides by Inuit (n = 120) that occurred between January 1, 2003, and December 31, 2006, in Nunavut, and all families provided written consent.

Control subjects (n = 120) were selected from the Nunavut Health Care Registration File. For each suicide, we randomly selected a community-matched control subject from a list of 5 sex-matched people who were born immediately before or after the person who died by suicide.

Procedures

We used a proxy-based interview procedure, also known as the psychological autopsy method, as the main source of information in our study. This method has established reliability and has been widely used in previous studies.^16,17 To ensure comparability between groups, information on control subjects was also obtained by means of proxy-based interviews. Overall, 498 interviews were conducted. The number of informants necessary to complete the interview was determined by the interviewer, based on the quality of the interviews and the amount of information gathered. For subjects, 279 interviews were conducted, whereas for control subjects 219 interviews were carried out. All interviews were conducted in English or Inuktitut according to the respondent’s preference. Medical charts, coroner’s notes, and criminal records were also systematically reviewed.

After the interviews and the review of medical and criminal records, the interviewers elaborated a detailed clinical vignette, with clinical and developmental histories. This vignette, which was stripped of any reference to group membership, was reviewed by a panel of clinician-scientists to generate consensus psychiatric diagnoses based on DSM, Fourth Edition, criteria.^18,19 A detailed description of the methodological procedures of this study has been published elsewhere.¹³

Measures

We considered risk factors from 3 domains: sociodemographical, psychopathology, and impulsive– aggressive behaviours. Sociodemographic variables were assessed using a standardized instrument that was culturally adapted to Nunavut.¹³ We also investigated the occurrence of childhood maltreatment, including physical, sexual, and (or) psychological abuse, using an abbreviated form of the Childhood Experience of Care and Abuse interview.²⁰ Information collected on history of maltreatment and legal problems was complemented with information collected from medical charts and criminal records. Family history was collected using an adapted version of the Family Interview for Genetic Studies, a semi-structured instrument that has been widely used in studies of psychiatric disorders.²¹

We assessed psychopathology using instruments adapted for proxy-based interviews. The Structured Clinical Interviews for DSM Axis I and II disorders^18,19 were used to elicit DSM-IV Axis I and personality disorders. This methodology has been extensively used in previous psychological autopsy studies and has been shown to be reliable and valid by our group and others.^22,23 We tested its validity in Nunavut by directly interviewing a random subsample of 30 control subjects and comparing the results with those obtained through informants. Kappa coefficients were 0.84 for mood disorders, 0.83 for personality disorders, 0.79 for schizophrenia, and 0.57 for substance abuse.

Impulsive behaviours were assessed by the BIS-11, and aggressive traits were assessed with the Brown–Goodwin History of Aggression.^24,25

Statistical Analyses

Statistical analyses were performed using SPSS, version 19.0.²⁶ All continuous variables were tested for normal distribution using Q-Q plot visual inspection, evaluation of skewness and kurtosis, in addition to the Kolmogorov– Smirnov test. Chi-square Fischer exact tests were used for univariate analyses of categorical variables, and Mann– Whitney U tests for continuous variables that failed to show normal distribution. Student t tests were used for univariate analyses of normally distributed continuous variables. Logistic regressions were subsequently used for multivariate analyses, controlling for potential confounders. VIF estimates greater than 5 were considered indicative of multicollinearity, and therefore the model was rejected.

Results

All Inuit deaths by suicide occurring between January 1, 2003, and December 31, 2006, and control subjects were enrolled in the study. To ensure collection of complete data and to minimize attrition rates, communities were visited several times. A total of 498 interviews were carried out (mean of 2.3 for subjects and 1.8 for control subjects). The mean age of the suicide subjects was 23.41 years (SD 9.13), while the mean absolute difference between subjects’ and control subjects’ birthdates was 59.27 days (SD 72.26), with a minimum difference of 1 and a maximum of 410 days.

Table 1 describes the sociodemographic risk factors across groups. Among the suicide subjects, 99 (82.5%) were male and 21 (17.5%) were female. Ninety-six of the subjects died by hanging (80.0%), 19 by gunshot (15.9%), 4 by stabbing (3.3%), and 1 by overdose (0.8%). Compared with control subjects, subjects were more likely to be single (OR 1.82; 95% CI 1.10 to 3.08), and to have had legal problems (OR 2.51; 95% CI 1.46 to 4.30). Subjects were less likely to have been working (OR 0.27; 95% CI 0.12 to 0.59), to have been a student (OR 0.48; 95% CI 0.22 to 0.98), and to have achieved higher education levels (OR 0.30; 95% CI 0.13 to 0.69 for junior high; and OR 0.25; 95% CI 0.12 to 0.52 for high school or higher).

Table 1.

Sociodemographic characteristics of the study sample

Sociodemographic characteristic	Suicide subjects n = 120 n (%)	Control subjects n = 120 n (%)	OR (95%CI)
Sex, male	99 (82.5)	99 (82.5)	n/a
Marital status
Married or common law	49 (40.8)	68 (56.6)	1.0
Single	63 (52.5)	45 (37.5)	1.82 (1.10 to 3.08)
Education level
<7 years	37 (30.8)	13 (10.8)	1.0
Junior high	24 (20.0)	28 (23.3)	0.30 (0.13 to 0.69)
≥High school	57 (47.5)	78 (65.0)	0.25 (0.12 to 0.52)
Occupation
No occupation	54 (45.0)	48 (40.0)	1.0
Worker	38 (31.6)	60 (50.0)	0.27 (0.12 to 0.59)
Student	12 (10.0)	28 (23.3)	0.48 (0.22 to 0.98)
Adopted, yes (%)	36 (30.0)	31 (25.8)	1.12 (0.69 to 2.14)
Legal history, yes (%)a	63 (52.5)	40 (33.3)	2.51 (1.46 to 4.30)
Number of people living in the household, mean (SD)	5.58 (2.09)	5.23 (2.60)	P = 0.27

^aDefined as having had charges pressed against the individual

n/a = not applicable

Subjects were also more likely to have been a victim of abuse than control subjects (57 and 33, respectively; OR 2.38; 95% CI 1.39 to 4.08). Sexual abuse during childhood was reported more frequently for subjects than for control subjects (15.83% and 6.66%, respectively; OR 2.63; 95% CI 1.10 to 6.27); while histories of physical abuse and psychological abuse were also more commonly reported among subjects (21.6% and 13.3%, and 20.0% and 10.8%, respectively), but this did not reach statistical significance (OR 1.79; 95% CI 0.90 to 3.55, and OR 2.05; 95% CI 0.99 to 4.26, respectively).

Subjects had an average of 1.61 previous suicide attempts (SD 1.36) per all subjects (subjects and control subjects), whereas control subjects had an average of 0.30 (SD 0.66) per subject (P < 0.001). Among control subjects, 26 (22.4%) had attempted suicide at least once. Among the subjects, 80 (66.6%) died from the first suicide attempt, 21 (17.5%) had one previous suicide attempt, 12 (10%) attempted suicide twice, and 7 (5.83%) had 3 or more previous suicide attempts. The mean time difference between the first suicide attempt and death was 1.81 years (SD 4.78).

Family history of psychopathology and suicide completion was more prevalent among first-degree relatives of subjects, compared with control subjects. Among subjects and control subjects, 62.2% and 23.0%, respectively, had at least one first-degree relative affected by MDD (χ² = 9.63, df = 1, P = 0.002), and 29.5% and 11.5%, respectively, had familial history of suicide completion (χ² = 5.41, df = 1, P = 0.02). However, the prevalence of suicide attempts among relatives was comparable between groups (27.3% of subjects and 24.6% of control subjects, χ² = 0.96, df = 1, P = 0.46). Abuse of, or dependence on, alcohol and drugs were also more prevalent among first-degree relatives of subjects, but did not reach statistical significance.

The number of first-degree relatives affected by at least one psychiatric disorder was significantly higher among subjects than control subjects. Table 2 describes the mean number of affected individuals per family.

Table 2.

Mean count of first-degree relatives with psychiatric disorders or suicidal behaviour across groups

Variable	Suicide subjects Mean (SD)	Control subjects Mean (SD)	Mann–Whitney U test	P
Major depressive disorder	0.48 (1.0)	0.15 (0.40)	2.07	0.04
Alcohol abuse or dependence	1.64 (2.33)	0.67 (0.85)	2.61	0.01
Drug abuse or dependence	1.07 (1.43)	0.61 (1.0)	1.93	0.05
Suicide completion	0.41 (0.72)	0.16 (0.48)	1.94	0.05
Suicide attempts	0.43 (0.78)	0.33 (0.65)	0.73	0.46

Six-month and lifetime prevalence of mental illnesses were significantly higher among subjects, compared with control subjects (6-month 85.0% and 43.3%, χ² = 45.30, df = 1, P < 0.001; lifetime 85.8% and 63.3%, χ² = 16.02, df = 1, P < 0.001). Univariate analyses showed that subjects had a higher risk of meeting criteria for MDD (OR 13.00; 95% CI 6.20 to 27.25 in the last 6 months, and OR 4.87; 95% CI 2.79 to 8.49 for lifetime), for abuse and (or) dependence of cannabis (OR 3.96; 95% CI 2.29 to 6.84 in the last 6 months, and OR 2.59; 95% CI 1.54 to 4.36 lifetime), and for abuse and (or) dependence of alcohol in the last 6 months (OR 2.90; 95% CI 1.59 to 5.24) but not lifetime (OR 1.27; 95% CI 0.76 to 2.12). Table 3 describes prevalence and unadjusted odds ratios for individual psychiatric diagnoses. Subjects had an average of 1.65 Axis I diagnoses (SD 1.03), compared with 0.61 (SD 0.82) for control subjects (z = −7.72, P < 0.001) in the last 6 months. Among control subjects, 68 had no Axis I diagnoses, 34 had only 1 diagnosis, and 18 had 2 or more diagnoses. Among subjects, 18 had no Axis I diagnoses, 32 had 1 diagnosis, 47 had 2 Axis I conditions, and 23 had 3 or more diagnoses. Comorbidity patterns were also different between groups for lifetime Axis I diagnoses (mean 1.84 [SD1.14] and 1.14 [SD1.10], respectively; z = −4.668, P < 0.001).

Table 3.

Unadjusted odds ratio of Axis I and II psychiatric diagnoses in subjects and control subjects

Axis I diagnoses	Suicide subjects n = 120 n (%)	Control subjects n = 120 n (%)	OR (95%CI)
Past 6 months
Major depressive disorder	65 (54.16)	10 (8.3)	13.00 (6.20 to 27.25)
Abuse or dependence of Cannabis	69 (57.5)	31 (25.8)	3.96 (2.29 to 6.84)
Abuse or dependence of alcohol	45 (37.5)	21 (17.5)	2.90 (1.59 to 5.24)
Schizophrenia	6 (5.0)	3 (2.5)	2.05 (0.44 to 10.64)
Obsessive–compulsive disorder	1 (0.8)	1 (0.8)	1.0
Pathological gambling	2 (1.6)	2 (1.6)	1.0
Posttraumatic stress disorder	3 (2.5)	3 (2.5)	1.0
Bipolar affective disorder type I	1 (0.8)	0	n/a
Lifetime
Major depressive disorder	73 (60.8)	29 (24.6)	4.87 (2.79 to 8.49)
Abuse or dependence of cannabis	71 (59.1)	43 (35.8)	2.59 (1.54 to 4.36)
Abuse or dependence of alcohol	53 (44.1)	46 (28.3)	1.27 (0.76 to 2.12)
Schizophrenia	6 (5.0)	3 (2.5)	2.05 (0.44 to 10.64)
Obsessive–compulsive disorder	1 (0.8)	1 (0.8)	1.0
Pathological gambling	2 (1.6)	2 (1.6)	1.0
Posttraumatic stress disorder	3 (2.5)	3 (2.5)	1.0
Bipolar affective disorder type I	1 (0.8)	0	n/a
Axis II diagnoses	Subjects n = 112 n (%)	Control subjects n = 84 n (%)	OR (95%CI)
Avoidant personality disorder	11 (9.83)	10 (14.28)	0.90 (0.33 to 2.31)
Borderline personality disorder	23 (20.53)	4 (4.76)	10.18 (3.34 to 30.80)
Antisocial personality disorder	17 (15.17)	9 (10.71)	2.90 (1.23 to 6.89)
Conduct disorder	18 (16.07)	4 (4.76)	2.90 (1.59 to 5.24)
Obsessive–compulsive personality disorder	2 (1.78)	2 (2.38)	2.05 (0.44 to 10.64)
Depressive personality disorder	2 (1.78)	1 (1.19)	1.49 (0.13 to 16.77)
Passive–aggressive personality disorder	0 (0)	1 (1.19)	n/a

n/a = not applicable

Concerning personality disorders, subjects were more likely to be diagnosed with cluster B personality disorders, but not cluster A or C. Borderline personality disorder was diagnosed in 23 subjects and 4 control subjects (20.53% and 4.76%, respectively; OR 10.18; 95% CI 3.34 to 30.80). Similarly, antisocial personality disorder was more prevalent among subjects than control subjects (15.17% and 10.71%, respectively; OR 2.90; 95% CI 1.23 to 6.89), and consistently, conduct disorder was also more common among subjects than control subjects (16.07% and 4.76%, respectively; OR 2.90; 95% CI 1.59 to 5.24). Table 4 describes the frequency and univariate analyses of the personality diagnoses observed in this study. Subjects presented an average of 0.89 (SD 1.08) personality diagnoses, whereas control subjects had an average of 0.29 (SD 0.60) (z = −4.546, P < 0.001).

Table 4.

Logistic regression model of risk factors for suicide

Variable	Adjusted OR (95%CI)
Marital statusa	1.05 (0.17 to 6.41)
Occupational statusb	1.29 (0.31 to 5.43)
Impulsivity levels (BIS-11 scores)	1.10 (0.97 to 1.25)
Aggressiveness (BGHA scores)	1.14 (1.06 to 1.24)
Major depressive disorder	75.72 (10.19 to 562.31)
Borderline personality disorder	1.15 (0.46 to 28.64)
Antisocial personality disorder	12.17 (0.37 to 301.04)
Conduct disorder	2.25 (0.46 to 111.1)
Abuse or dependence of cannabis	1.56 (0.34 to 7.12)
Abuse or dependence of alcohol	1.07 (0.19 to 5.99)
Victim of abuse	1.07 (0.20 to 5.61)

^aMarital status 0 = married or living with common-law partner, 1 = single or separated

^bOccupational status 0 = unemployed, 1 = student or worker

BGHA = Brown-Goodwin History of Aggression;

BIS-11 = Barratt Impulsivity Scale (second version)

Consistent with cluster B personality differences, impulsivity was significantly higher in suicide subjects. BIS-11 scores were 76.04 (SD 5.93) for subjects and 71.04 (SD 6.35) for control subjects (t = 6.08, df = 1, P < 0.001). Similarly, levels of aggressive behaviours were higher among subjects (mean 58.05 [SD 18.29] for subjects and 39.63 [SD 8.59] for control subjects, Mann–Whitney U = 1672.5, z = −6.97, P < 0.001).

Logistic regression was employed to identify independent risk factors for suicide among Inuit. We included Axis I diagnoses in the last 6 months (rather than lifetime) in the model, as they had a temporal association with the suicide. Marital status, educational level, and occupation were dichotomized (married or living with common-law partner, compared with single or divorced; less than 7 years of schooling, compared with 7 or more years of schooling; and worker or student, compared with without occupation, respectively). The model explained a large proportion of the variance in the outcome (Nagelkerke R²_n = 0.778), had good prediction power (overall 91%), and presented satisfactory goodness-of-fit (Hosmer–Lemeshow test P = 0.76). All independent variables presented VIF lower than 2. The presence of MDD and aggressive behaviours predicted suicide in the regression model. Adjusted odds ratios are presented in Table 4.

Discussion

Our study is the first to systematically assess individual-level risk factors for death by suicide in indigenous people using standardized psychiatric assessments and psychological autopsies. It is also the first project to include a representative sample of all deaths by suicide in an indigenous population. We included a control group matched by age, sex, and community of origin to ensure that the environmental conditions and historical context during the subjects’ upbringing were comparable across groups. This is particularly relevant in the case of small and remote communities, where local events may have a strong impact on many people’s lives.

Our results indicate that both psychopathology and early life adversity are associated with death by suicide among Inuit. Probably the most striking finding from our study is that, contrary to expectations,²⁷ and some studies in non-Western populations, such as recent studies from Asia,^11,28 clinical risk factors for suicide among Inuit were not different from those observed in studies with the general population, at least at the individual level.⁹ MDD and levels of aggressive behaviours proved to be the most significant risk factors for suicide completion in our study. Moreover, family history of suicide, childhood sexual abuse, impulsivity, and comorbidity with substance abuse were also associated with suicide completion in the univariate analyses. These are all well-established risk factors for suicide and consistent with results from psychological autopsy studies exploring risk factors for suicide in various Western populations.⁹

Our study also found a relatively high prevalence of psychiatric disorders among control subjects. For instance, almost one-quarter of all control subjects had a history of MDD and more than one-third had cannabis use disorder. A history of suicidal behaviour was also frequent among control subjects. While lower than those found among subjects, these rates are all significantly higher than rates in the general population of Canada.²⁹ These high rates are probably associated with elevated levels of psychological distress among Inuit in the region.³⁰

While suicide is recognized as an important public health problem in Nunavut and many other Indigenous populations in Canada and around the world, there is little empirically based information to guide prevention programs. Our study suggests several strategies for suicide prevention and intervention for the Inuit population. Given the high rates of potentially treatable mental health problems, improved access to mental health services should be implemented across Nunavut. However, the high rates of psychopathology among control subjects points to the need for broader mental health promotion programs, particularly addressed to youth.

Our findings are consistent with an understanding of suicide from a developmental perspective. The life histories collected in this study suggest that the trajectory that leads to suicide may begin in childhood, and that both distal and proximal intervention and prevention strategies should be included.

Distal Strategies

The excess of history of early life adversity among subjects is consistent with the increased impulsive–aggressive traits observed.³¹ Interventions targeting parenting styles and helping foster parenting skills in young families have produced promising results,³² decreasing rates of maladaptive behaviours and long-term mental health outcomes. Moreover, early school programs and specific social interventions for youth to develop coping skills could also be implemented. Indigenous suicide prevention efforts should consider such intervention programs in their long-term strategies.

Proximal Strategies

Programs on mental health literacy, coping strategies, and psychopathology have been used successfully to promote awareness in youth about suicide risk factors and to develop coping mechanisms when under stress.³³ These promising strategies can be adopted for Inuit communities, both to fit culture and the geographic setting of small, remote communities in Nunavut, where web-based and other forms of e-health intervention and (or) prevention strategies may be particularly useful.

Our findings also indicate future research directions. There is a need for further study of the developmental processes that may lead to suicidality. Moreover, the specific triggers that provoke the transition from suicidal ideation to attempts are not adequately understood in Indigenous contexts. Further research is needed to identify effective prevention strategies. Implementation of programs and strategies to address risk factors (such as childhood adversities, mental health disorders, and social challenges) should be coupled with research to assess the process of implementation and the outcomes of the interventions.

Limitations and Strengths

Our study is correlational and cannot make inferences about causal relationships. Our study focused exclusively on individual risk factors, while other factors, such as those acting at the social level, are likely to explain important variations in suicide rates observed during the last decades among Inuit and other Indigenous peoples. For instance, negative outcomes, such as low self-esteem, drug misuse, sexual abuse, and violence, have been reported among the descendants of people who attended residential schools.^34,35 While Indigenous peoples share many of the social stressors and putative risk factors for suicide, they also have important differences. Our study focused exclusively on an Inuit population, and it is unclear how generalizable the observations are to other Indigenous communities.

Our study also presents unique strengths. It is the first to enrol the totality of people who died by suicide in one region, which ensures its representativeness. Moreover, we used valid and reliable measures to allow for comparison with previous studies. Although necessarily retrospective, the psychological autopsies included multiple sources of information and were blindly reviewed.

Conclusion

Among Inuit in Nunavut, as in the general population, mental health disorders and early life adversity are major risk factors for suicide. In addition, control subjects also presented high rates of psychiatric disorders. Consequently, mental health interventions aiming at treating psychiatric conditions and broader population-level mental health promotion to improve family environments should both be components of any suicide prevention strategy in Inuit communities.

Abbreviations

BIS-11

Barratt Impulsivity Scale (second version)

DSM

Diagnostic and Statistical Manual of Mental Disorders

MDD

major depressive disorder

VIF

variance inflation factor