A Systematic Review of the Antecedents and Prevalence of Suicide, Self-Harm and Suicide Ideation in Australian Aboriginal and Torres Strait Islander Youth

Suicide and self-harm represent serious global health problems and appear to be especially elevated amongst indigenous minority groups, and particularly amongst young people (aged 24 years or younger). This systematic review investigates for the first time the antecedents and prevalence of suicide, self-harm and suicide ideation among Australian Aboriginal and Torres Strait Islander youth. Web of Science, PubMed, PsychINFO, CINAHL databases and grey literature were searched from earliest records to April 2019 for eligible articles. Twenty-two empirical articles met the inclusion criteria. The data confirmed that indigenous youth in Australia have elevated rates of suicide, self-harm and suicidal ideation relative to the nonindigenous population. Risk factors included being incarcerated, substance use and greater social and emotional distress. Notably, though, information on predictors of suicide and self-harm remains scarce. The findings support and justify the increasing implementation of public health programs specifically aimed at tackling this crisis. Based on the review findings, we argued that Aboriginal communities are best positioned to identify and understand the antecedents of youth self-harm, suicide ideation and suicide, and to take the lead in the development of more effective mental health preventive strategies and public policies within their communities.


Introduction
Suicide is the cause of over 800,000 deaths annually around the globe [1]. The World Health Organization (WHO) has urged that suicide prevention be given a higher priority on the global public health agenda [1]. Suicide attempts and other acts of self-harm are even more prevalent and are associated with an elevated risk of eventual death by suicide, as well as reduced lifespan more generally [2][3][4]. Here we use the term self-harm to encompass both suicide attempts and non-suicidal self-injury ((NSSI) e.g., self-cutting as a means to manage or cope with difficult emotional states) [5][6][7][8]. Self-harm and suicidal ideation are markers of considerable emotional distress, often associated with other psychological difficulties, and so remain important clinical outcomes in their own right [9][10][11][12]. It has been recognised that suicide rates are often elevated amongst Indigenous populations, including Australian Aboriginal and Torres Strait Islander peoples [1,13]. Young people can be especially vulnerable to some of the factors that contribute to risk of suicide and self-harm. Data on the prevalence of suicide, self-harm, and suicidal ideation in these populations is important in determining the level of ("Aborigin*" OR "Indigenous" OR "Torres Strait Islander") AND (Australia*) AND ("adolescen*" OR "teen*" OR "youth" OR "young people" OR "child*").
Initially, three reviewers (K.C., S.D. and C.Mc.) independently screened the titles and abstracts of all identified articles. They then further screened the selected full articles with disagreements arbitrated by two reviewers (J.M.D. and P.J.T.). Reference lists of all articles were checked for additional relevant publications. Authors of the selected review manuscripts were also contacted to see if they had any other possibly eligible papers for inclusion in the review. One additional paper was returned but deemed ineligible according to the criteria. Where identified articles discussed findings concerning suicide and self-harm but specific statistics relating to either Aboriginal or Torres Strait Islander young people were not provided, the corresponding authors of these papers were approached and additional data requested (n = 11). We received nine responses and no new relevant information was provided. Data, including study characteristics (e.g., authors, date of publication), study methodology (e.g., design, measures used), participant characteristics and study results (e.g., prevalence or incidence data and associations between variables of interest) were extracted using a spreadsheet template by the researchers to aid synthesis.

Risk of Bias
Included studies were assessed using the Agency for Healthcare Research and Quality methodological quality assessment tool [24]. This tool has been designed to evaluate risk of bias in observational research and was adapted (as in previous studies) for the specific context of this review [25]. The risk of bias assessment for each study is reported in Table 1. We also adopted a narrative synthesis of the risk of bias assessment to report common areas of weaknesses across studies.

Results
A flow chart of the literature screening is presented in Figure 1. Overall, 22 papers were identified for inclusion in the review. Due to the considerable heterogeneity in study design and measures of suicide, self-harm and suicidal ideation (e.g., self-report, hospital morbidity data and coronial data), aggregation of effect sizes was limited by high heterogeneity and low precision, so meta-analysis was not used here. The results were therefore synthesised narratively. Studies were grouped into five (not mutually exclusive) categories, including studies that evaluated the prevalence of suicide among indigenous youth via analysis of coronial data (k = 8), [26][27][28][29][30][31][32][33]; studies that evaluated the prevalence of self-harm and attempted suicide via analysis of hospital admissions records (k = 3), [34][35][36]; studies that evaluated the prevalence of suicide, self-harm, or suicide ideation among Indigenous youth in the community (k = 7), [37][38][39][40][41][42][43]; studies that involved samples of incarcerated Indigenous youth (k = 4), [44][45][46][47]; and studies that evaluated risk factors associated with suicide, self-harm, and suicidal ideation among Indigenous youth (k = 5), [33,[38][39][40][41]. The characteristics of the studies included in this review are summarised in Table 1.
Although several studies did not provide a justification for sample sizes (e.g., power calculations), these studies typically relied on large datasets (10 studies with n > 300) where problems associated with low power are less likely. For the most part, blinding of researchers to participants' background was not relevant because studies relied on secondary analysis of existing datasets, where rater bias is unlikely to have had an impact (e.g., detection bias associated with an interviewer being more likely to prompt in questions about self-harm if participants have an Aboriginal background). In those studies that relied on interviews or direct assessments to collect study data such biases are possible. However, blinding to ethnicity in studies involving face-to-face interviews may not be realistic. For those studies focused on determining predictors or correlates of suicide, self-harm or suicidal ideation, five studies did not attempt to control for potentially confounding variables. Parameter estimates for these particular studies may be biased as relevant confounders were not taken into account. While most studies used validated measures for determining suicide and self-harm, a common methodological issue associated with the analysis of hospital records was the lack of differentiation between self-harm with suicidal intent and non-suicidal self-injury. For the most part, however, the overall quality of the research conducted was of a good standard. In those studies that relied on interviews or direct assessments to collect study data such biases are possible. However, blinding to ethnicity in studies involving face-to-face interviews may not be realistic. For those studies focused on determining predictors or correlates of suicide, self-harm or suicidal ideation, five studies did not attempt to control for potentially confounding variables. Parameter estimates for these particular studies may be biased as relevant confounders were not taken into account. While most studies used validated measures for determining suicide and selfharm, a common methodological issue associated with the analysis of hospital records was the lack of differentiation between self-harm with suicidal intent and non-suicidal self-injury. For the most part, however, the overall quality of the research conducted was of a good standard. Eleven studies provided prevalence data estimates for suicide rates among Indigenous relative to nonindigenous youth (see Table 3). Relative to nonindigenous young people, age-standardised   Eleven studies provided prevalence data estimates for suicide rates among Indigenous relative to nonindigenous youth (see Table 3). Relative to nonindigenous young people, age-standardised prevalence rates were higher for Indigenous populations for both children (Indigenous: 1.5 to 4.61 per 100,000 vs. nonindigenous: 0 to 0.48 per 100,000) and adolescents (Indigenous: 35.6 to 57.50 per 100,000 vs. nonindigenous: 11.7 to 14.33 per 100,000) [29,30]. Studies reported that Indigenous children (15-years of age or less) were between 10 and 14 times more likely to die by suicide, and Indigenous youth (15-24 years) were between four and 14 times more likely to die by suicide than their nonindigenous peers [29,32,33]. Further, the risk of dying by suicide was more than twofold greater for indigenous children than for indigenous adolescents [32].
A recent study identified increasing rates of hospital admissions for intentional self-harm and suicide ideation among Indigenous children living in the Northern Territory, Australia [35]. However, all hospital admissions associated with ICD-10 codes for intentional self-harm and suicidal ideation were analysed as one variable, thus limiting our understanding of the prevalence of self-harm and suicidal ideation, respectively, among this population. These same authors reported that the average annual change in the number of hospital admissions was much greater among Indigenous children (23.5%) than Indigenous youth (11%) and older Indigenous cohorts (25-54 years: 8-13% average annual change).
Time periods over which prevalence rates were estimated however varied between studies (two weeks (k = 1), 12 months (k = 1), lifetime prevalence (k = 2) and not reported (k = 3), thus limiting comparisons between studies. One study [37] indicated that almost half (46% n = 450) of all calls made to a telephone counselling service by Aboriginal young people related to suicidal ideation or suicide related concerns. Studies also suggested that the prevalence of suicidal ideation was higher among Aboriginal youth who were incarcerated [47] compared to Aboriginal youth living in the community [39,40]. However, amongst incarcerated youth, two studies suggested no difference between Indigenous and nonindigenous detainees or those on remand with regards to suicidal ideation [44,46]. It is possible that ceiling effects account for the lack of differentiation regarding youth suicide ideation in these two studies, as incarcerated youth generally represent a high risk population.
Four studies evaluated the prevalence of self-harm among Indigenous youth [32,36,42,43]. However, cases with and without intent were not differentiated in three of these studies [32,36,43], and suicidal intent was reported in one study [42]. Self-harm was identified as a key issue among Aboriginal youth seeking telephone counselling support, with more than half (59%, n = 450) of young Aboriginal callers seeking assistance for self-injury or self-harm related concerns [37]. Age-standardised rates of self-harm, estimated from hospital admissions records suggest that the prevalence of self-harm was much higher among Aboriginal youth (74.23 per 100,000) than for both non-Aboriginal Australian youth (29.18 per 100,000) and young UK born migrants (40.22 per 100,000) [36]. These estimates however were based on data collected between 1984 and 1993. There are also limitations associated with the use of hospital records when estimating the prevalence of self-harm among any population, as a substantial number of self-harm incidents occur that either do not require medical attention or for which medical attention was not sought [11]. Therefore, the rates of self-harm identified by Rock and Hallmayer [36] are likely to be under estimates of self-harm. However, a more recent study comparing three Indigenous communities in Queensland similarly reported a high incidence rate of 30.1 per 1000 for deliberate self-harm presentations among Indigenous young people aged 15-24 years, whereas for those aged 15 years or younger the incidence rate was significantly lower, 1.6 [43]. Lifetime prevalence of actual suicide attempts was found to be significantly higher among Aboriginal or Torres Strait Islander young people aged 18-24 years (14.7; 95% CI 8.0, 25.6) than nonindigenous young people in the same age group (6.3; 95% CI 5.0, 7.9), whereas no significant difference was reported among Indigenous and nonindigenous young people aged between 14 and 17 years (p = 0.8) [42]. Risk factors for suicide, self-harm and suicide ideation are presented in Table 4.  [37] Yes n/a n/a Partial Yes Partial n/a n/a Yes n/a n/a        [37] Not Investigated Fifty-nine percent of Aboriginal callers sought assistance for self-injury and/or self-harm concern.
Forty-six percent of Aboriginal callers reported suicide related concern and/or ideation 48% (n = 71) of Indigenous deaths by suicide were reported among those aged <20 years.

Not Investigated Not Investigated
Clayer & Czechowicz (1991) [28] 32.6% (n = 15) of Aboriginal suicides occurred among 10-19 year olds compared to 8.4% (n = 110) of non-Aboriginal young people of the same age range.  [46] No sig difference in prevalence of suicide behaviour between Aboriginal and non-Aboriginal adolescents on remand (p-values not reported).

Not Investigated
No significant difference in prevalence of suicidal ideation between Aboriginal and non-Aboriginal adolescents on remand (p-values not reported).  Young Aboriginal females reported sig higher levels of suicidal ideation than young Aboriginal males (β = 0.49, 95% CI = 0.25-0.74). Aboriginal youth experienced racial discrimination were at a significantly higher risk of suicide ideation than those who did not (β = 0.34, 95% CI = 0.08-0.60). No significant difference in level of suicidal ideation among Aboriginal youth living in regional and remote areas. No significant difference in level of suicide ideation among Aboriginal youth who were employed vs.  Racial discrimination associated with significant increased risk of suicidal ideation (OR, 2.32 (95% CI, 1.25-4.00; p = 0.001)).

Not Investigated Not Investigated
Zubrick et al. (2011) [41] Aboriginal children whose birth mother had died were at a sig higher risk for parent-reported suicide attempts (OR = 7.0, 95% CI = 1.6-31.1 p value not reported).
No significant difference between Aboriginal children who lived with their birth mother or whose birth mother had died with regards to self-harm.
Aboriginal children whose birth mother had died were at a sig higher risk for parent-reported suicidal ideation (OR = 2.6, 95% CI = 1.2-5.7; p value not reported). Several risk factors for suicide, self-harm and suicide ideation were identified in the systematic review. The findings were all based on cross-sectional designs (see Table 4). Evidence to support sex as a risk factor for suicide and suicidal ideation were mixed. Jamieson and colleagues [38] reported higher levels of suicidal ideation among Indigenous females, relative to Indigenous males, whereas Luke and colleagues [39] reported no significant gender differences. There were however important demographic differences between these studies that may account for the discrepant findings concerning suicide ideation. Luke and colleagues sampled young Indigenous people living in a metropolitan area where a very low proportion of the population identified as Indigenous [39]. In contrast, Jamieson and colleagues (2011) sampled young Indigenous people residing in the Northern Territory where 29.8% of the population identify as Indigenous. Further, another study did not find significant gender differences between young Indigenous males and females in the reported incidence of self-harm [43]; nor was sex associated with increased rates of attempted suicide [39] or death by suicide [32] among Indigenous young people (k = 2). However, one study did report higher suicide rates among 5-14 year-old Indigenous males (5-14 years: male= 5.57 per 100,000; female = 3.6 per 100,000) and substantially higher suicide rates among 15-24-year-old males (males = 91.96 per 100,000; females = 22.74 per 100,000), relative to indigenous female [29]. Although Soole and colleagues [32] and De Leo and colleagues [29] extracted data from the Queensland Suicide Register, De Leo and colleagues estimated suicide rates over a much longer time period (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007).
Geographic location was identified as an important risk factor for suicide [33,38]. Particularly high suicide rates were reported among Indigenous children living in remote areas of Australia (Indigenous: 33.75 per 100,000; nonindigenous: 0 per 100,000) [33].
Indigenous children in regional areas were also at a significantly higher risk of suicide than nonindigenous children (Indigenous: 9.5 per 100,000 vs. nonindigenous: 1.4 per 100,000), whereas suicide rates did not differ significantly between Indigenous and nonindigenous children living in metropolitan areas (Indigenous = 0 per 100,000; nonindigenous = 0.56 per 100,000) [33]. Indigenous children were significantly more likely to be residing outside of the parental home prior to suicide and to die by suicide outside of their family home compared to nonindigenous children (p = 0.03) [33].
The experience of racial discrimination was identified as a significant risk factor for suicidal ideation among young Aboriginals (β = 0.34) [38,40], particularly among those living in the Northern Territory. One study conducted with Aboriginal youth living in a metropolitan region of Australia (Victoria) identified a trend toward a significant association between lack of cultural connection and suicidal ideation (p = 0.06) [39].
Only one study reviewed considered a wide range of psychosocial and cultural factors in relation to risk of suicidal ideation and suicide behaviours [39]. An increased risk of suicidal ideation was associated with emotional distress (OR = 7.6) and social distress (OR = 2.0); aggregate variable encompassed: no friends to talk to, few friends, parents with substance abuse problems, physical abuse and previous youth detention. An increased risk of suicide attempts was also associated with emotional distress (OR = 2.5) and two aggregate social distress variables (OR= 2.5-3.2; encompassing few friendships, no adults to talk to, parents with substance problems, parents not living together, physical abuse, previous youth detention, and low importance of Koori Aboriginal values). Increases in putative risk behaviours (no participation in sport, smoking, heavy drinking, marijuana use) were not associated with increased risk of suicidal ideation; however, there was a trend towards a significant increased risk of suicide attempt (OR = 1.8). Similarly, Jamieson and colleagues [38] reported that risk of suicidal ideation was not related to alcohol or other drug use, but these authors did not investigate risk of suicide attempt. One study investigated the impact of the bereavement, specifically, the death of the birth mother on risk of suicidal ideation and suicide attempts among young Indigenous children [41]. These authors identified an increased risk of suicidal ideation (OR = 2.6) and suicide attempt (OR = 7.0), but not deliberate self-harm (non-suicidal self-injury). This study relied on parent and carer reports, however, which may have underidentified the true rate of self-harm and suicidal ideation.

Discussion
The current review aimed to synthesise the empirical literature on both the prevalence and antecedents of suicide, self-harm and suicide ideation among Australian Aboriginal and Torres Strait Islander youth. In summary, our main findings were (1) the age standardised suicide rate for Indigenous youth, and Indigenous children in particular, was substantially higher than nonindigenous counterparts; (2) prevalence of self-harm and suicide ideation were higher among Indigenous youth than nonindigenous youth; (3) greater risk of suicidal ideation among Indigenous youth was associated with being incarcerated, experience of racial discrimination, and emotional and social distress, but not substance use; (4) living in regional and remote areas was associated with greater risk of suicide, alongside substance use, being incarcerated and high levels of social and emotional distress were identified as risk factors for suicide; (5) evidence that sex represented a risk factor for suicide and suicidal ideation was inconsistent and there was no evidence available for sex as a risk factor for self-harm; (6) data on prevalence and antecedent risk factors for self-harm among Indigenous youth was limited, although available evidence suggests increased prevalence of self-harm among Indigenous youth compared to nonindigenous youth; and (7) there was a surprising lack of empirical research literature on the antecedent risk factors for suicide, self-harm and suicidal ideation among Indigenous youth.
The current review highlights substantially increased suicide rates among young Aboriginal and Torres Strait Islander youth, relative to nonindigenous Australian youth, thus confirming the emergence of a suicide crisis among young Indigenous populations. We argue the term 'crisis' is appropriate, since young Indigenous Australians are not only dying by suicide at significantly higher rates than their nonindigenous peers but they are dying by suicide at an increasingly younger age, particularly in remote regional areas. Suicide rates for young Indigenous Australians aged 15-24 years (39.4 per 100,000) are far higher than the national rate for young people (10.7 per 100,000), [13] and the global suicide rate among young adults 15−29 years, which accounts for 8.5% per 100,000 of all deaths [1]. Prevalence data regarding self-harm in Indigenous Australian youth was more limited. Self-harm is a more difficult phenomenon to monitor, especially at the less medically serious end of the spectrum where individuals may engage in self-harm privately with little contact with social or health services. We argue this crisis extends beyond mental health and encompasses wider social, cultural and emotional factors.
Although much of the available literature has focused on the prevalence of suicide and to a lesser extent suicidal ideation, there is a notable lack of empirical evidence on the antecedents of suicide, self-harm and suicidal ideation among young Australian Indigenous populations. Given that self-harm is a significant predictor of suicide [2][3][4], it is particularly pressing to investigate antecedent risk factors in order to develop more effective prevention and intervention strategies. In particular, it is imperative to identify the key socio-cultural factors (e.g., discrimination and economic deprivation) and psychological factors (e.g., beliefs, thoughts, coping strategies and emotional states) that contribute to elevated risk of suicide and self-harm among Indigenous youth. While this review strongly suggests some candidate variables, including discrimination, social and emotional distress, evaluation of these predictors was limited and no studies examined whether they actually mediated the relationship between belonging to Aboriginal and Torres Strait Islander populations and the risk of suicide, self-harm or suicidal ideation.
Limitations associated with the studies reviewed deserve comment. The lack of longitudinal studies means that the temporal characteristics or direction of the relationship between risk factors and suicide or self-harm is unclear. Western conceptualisations of self-harm may not apply to Aboriginal and Torres Strait Islander populations, as the antecedent thoughts and behaviours that increase the risk of suicide among Indigenous populations may differ from that of the general nonindigenous population [21,[48][49][50][51][52]. Those studies that included self-report assessments of self-harm and suicidal ideation, were limited to a brief set of yes/no items (k = 3) or a single item assessment (k = 1). Further, assessment of suicidal ideation and/or self-harm was not adequately described (k = 3), e.g., the works by the authors of [39,40,46]. As such, a more detailed examination of the frequency and severity of these phenomena is precluded. Future research would benefit from the development of more culturally sensitive and appropriate measures to assess suicide risk.
The over representation of studies reporting morbidity and mortality data in the current review reflects the accessibility of this population level data. Indigenous suicides are often not effectively identified by coroners and therefore there may be significant under reporting [20]. Further, there is a surprising lack of qualitative research. The current review did not exclude qualitative methods from the search strategy, however, no qualitative research designs or assessments of suicide, self-harm or suicidal ideation among Aboriginal and Torres Strait Islander youth were identified. It could be argued that it may be more effective to undertake qualitative research in a culturally sensitive manner (e.g., narrative analysis) as these approaches seek to understand the 'insider' experience and complexity of a particular phenomenon. Meta-analysis would have helped to derive more precise estimates of prevalence and associations with risk factors, but was not possible due to substantial heterogeneity in study designs and outcomes and the likely nonindependence of studies relying on large national datasets.
Based on the present systematic review, the findings highlight some key avenues for future research and public health policy. Improvement in routine collection of self-harm information via hospital and mental health service providers, along with standardised reporting systems would allow for national level statistics to improve prevalence estimates. Large scale longitudinal studies would provide a better test of predictors of risk, for both suicidal ideation and self-harm among Aboriginal and Torres Strait Islanders. Health and social practitioners working with Aboriginal and Torres Strait Islander communities should be mindful of the elevated suicide and self-harm risk factors. It has been shown that psychosocial interventions can help reduce self-harm risk [53], but trials of culturally adapted interventions that specifically target self-harm in Aboriginal and Torres Strait Islander populations are sparse [54]. A trial of a self-help mobile phone application, designed with Aboriginal and Torres Strait Islander people, aimed at reducing suicide ideation is currently ongoing [55]. Given the high proportion of Indigenous people living in regional and remote areas, and the high incidence of youth suicide in such areas, a research agenda aimed at profiling the unique psychological, social and cultural needs of these communities in relation to suicide and self-harm is required. Research aimed at studying self-harm, suicide ideation and suicide among incarcerated youth awaits further investigation. Many Aboriginal and Torres Strait Islander communities are attempting to tackle the serious crisis of youth suicide [51,52]. Aboriginal communities are best positioned to identify their specific research questions to better understand the antecedents of youth self-harm, suicide ideation and suicide, and the development of more effective preventive strategies and public policies, within their specific communities. Given many of the risk factors identified in this review are social or societal in nature, including discrimination or environment, broader social policy initiatives may also be an important step in reducing self-harm and suicide. These may include initiatives to reduce discrimination, increase social cohesion, preserve culture and promote quality of life and self-determination in the community.

Conclusions
This review highlights the substantially elevated rates of suicide, self-harm and suicidal ideation amongst Aboriginal and Torres Strait Islander young people in Australia compared to nonindigenous young people. These elevated rates reflect the complex range of social, cultural and psychological adversities faced by the Aboriginal and Torres Strait Islander youth population in Australia, but further research is needed to delineate these factors. The findings strongly support and justify the increasing call for the implementation of public health programs specifically aimed at tackling the crisis of suicide and self-harm among Aboriginal and Torres Strait Islander young people [51,52,[56][57][58]. Such programs will be strengthened by a greater recognition of the specific social, cultural, psychological and intergenerational determinants of self-harm and suicide among Aboriginal and Torres Strait Islander young people. Aboriginal and Torres Strait Islander communities need to be at the forefront in shaping and implementing these programs to tackle this urgent crisis.