Long-term risk of repeat hospital admission involving self-harm by Aboriginal and non-Aboriginal people: A retrospective cohort study

Introduction

The Northern Territory (NT) of Australia has the highest rates of hospitalised self-harm [1] and deaths from suicide [2] of any jurisdiction in Australia. Self-harm is one of the strongest predictors of suicide [3], which makes it an important and urgent target for suicide prevention in the NT. Additionally, the repetition of self-harm is associated with further increasing the risk of suicide [4–6]. This has also been investigated in the NT in a cohort of hospital admissions involving self-harm and suicidal ideation, where the risk of suicide was found to increase by up to 30% for each repeat hospital admission involving self-harm [7]. Therefore, preventing repeat hospital admissions involving self-harm can contribute to reducing the risk of suicide in the NT.

Whilst there is evidence to inform measures that prevent the repetition of hospital-treated self-harm [8], its relevance to the unique context of the NT remains unclear. The NT has the highest proportion of Aboriginal and Torres Strait Islander (hereafter, respectfully referred to as Aboriginal) residents. Aboriginal people are the sovereign owners of and share collective and ancestral ties to the lands, waters, and skies that are intimately linked to their physical, psychological and spiritual wellbeing. They hold vital knowledge systems and expertise about health and wellbeing, but are grappling with the ongoing effects of colonisation contributing to health and social inequities and compromises in access to quality health care [9]. Aboriginal peoples continue to experience a higher risk of self-harm and suicide. A considered response is required which supports a self-determined cultural approach to address the complex systemic, social, historical, and political influences that are distinct from those affecting non-Aboriginal people in Australia [10]. However, only two recent cohort studies are known to have reported outcomes following hospitalisation involving self-harm, suicidal ideation, or both amongst Indigenous peoples from post-colonial countries similar to Australia. A cohort study from New Zealand found Māori were at higher risk compared to non-Māori of repeat presentation to ED involving self-harm, but this association was insignificant after adjusting for other demographic and clinical characteristics [11]. In a Canadian cohort study of children and adolescents hospitalised with self-harm the risk of repeat self-harm and suicide was higher amongst people who identified as American Indian or Alaska Native descent [12]. Whilst both studies suggest a greater risk of repeat self-harm exists amongst Indigenous peoples, they do not clarify the distinct risk factors that may need to be targeted for each of these population groups. Nor have these cohort studies considered suicidal ideation as an exposure, which has been identified as a potentially important expression of suicide risk amongst Aboriginal people in Australia [13,14].

Few longitudinal studies have highlighted risk factors for repeat hospitalisations involving self-harm amongst people initially hospitalised with suicidal ideation. In a study using Northern Ireland’s national self-harm registry surveillance data from episodes of hospitalisation involving both thoughts about and acts of self-harm, the risk of repeat hospital presentations involving self-harm was higher amongst individuals with an initial presentation involving suicidal ideation compared to self-harm [15]. Whilst the risk of repeat presentations involving self-harm was found to increase with the number of previous episodes of self-harm in a study from Spain, no differences in relative risk were observed between individuals presenting with self-harm and those presenting with suicidal ideation at the initial hospital presentation [16]. Further research is needed to bolster the mixed but limited evidence regarding the risk of self-harm following suicidal ideation in hospital settings.

In order to address these gaps in the evidence and inform efforts to reduce the population burden of self-harm in the NT a whole population retrospective cohort study was designed. Our aim with this study was to identify distinct risk factors amongst Aboriginal and non-Aboriginal people separately for repeat hospital admissions involving self-harm following an initial admission involving self-harm and/or suicidal ideation.

Methods

Results

During the study period, 4,391 individuals recorded an index hospital admission involving self-harm (n=2,956), suicidal ideation (n=1,292), or both (n=143). Just over half of the cohort (n=2,304; 52.5%) identified as Aboriginal, just under half (n=2,084; 47.5%) were female, and the median age at index admission was 30 years (IQR: 22-40). A more detailed summary of the demographic and clinical characteristics for Aboriginal and non-Aboriginal people is provided in the supporting information (see S1 Table and S2 Table).

During the follow-up period, 943 individuals were admitted on 1,985 occasions for self-harm (Aboriginal n=1,208; non-Aboriginal n=777). Of those with a repeat hospital admission involving self-harm, the number of episodes ranged from 1 to 28 (median: 1; IQR: 1-2). After accounting for censoring, 25.3% (95% CI: 23.8%-27.0%) of individuals in the study had experienced at least one repeat hospital admission involving self-harm (Aboriginal: 29.9%, 95% CI: 27.5%-32.4%; non-Aboriginal: 20.3%, 95% CI: 18.3%-22.4%). Figure 1 shows the survival probability (i.e., probability of not experiencing a repeat hospital admission involving self-harm) by Indigenous status illustrating a 40% higher risk of the outcome for Aboriginal compared to non-Aboriginal people over the course of the study (HR: 1.39; p < 0.001; 95% CI: 1.22-1.59). Table 1 provides a further summary of the distribution of absolute risk of repeat hospital admissions involving self-harm across demographic characteristics and type of suicidal behaviour at index admission for Aboriginal and non-Aboriginal people.

• Download figure
• Open in new tab

Figure 1

Probability of survival for Aboriginal and non-Aboriginal people following hospital admission involving self-harm and suicidal ideation, NT 2001-2013

In the multivariable modelling for both Aboriginal and non-Aboriginal people (Table 2), no strong evidence of substantial differences was observed in the relative risk of repeat hospital admission involving self-harm according to demographic characteristics, except for age in non-Aboriginal people. Compared to index admissions involving suicidal ideation alone, all methods of self-harm carried a higher risk of repeat hospital admission involving self-harm for Aboriginal people, whereas for non-Aboriginal people this was only the case for index admissions involving self-poisoning. Additionally, index admissions involving hanging by non-Aboriginal people had an approximately 2 times lower risk of repeat hospital admission involving self-harm compared to index admissions involving suicidal ideation. Diagnoses of substance use disorders, especially in relation to alcohol, were the most notable clinical risk factor associated with the outcome for both Aboriginal and non-Aboriginal people. For Aboriginal people, this association was found for diagnoses of alcohol use disorders in both index and previous hospital admissions, although the statistical support for the former was weak after adjusting for other factors. For non-Aboriginal people, diagnoses of alcohol use disorders at index admission and of substance use disorder in previous admissions were associated with repeat hospital admissions involving self-harm. Furthermore, diagnoses of severe mental disorders and personality disorders at index admission and diagnoses of common mental disorders in previous admissions were all independently associated with a higher risk of repetition for non-Aboriginal people. However, the pseudo R² statistics estimated from the best fitting models (R² : 0.097; R² : 0.109) suggests that very little of the variance in the outcome was explained by the available data.

Discussion

This is the first known cohort study to distinguish the risk of repeat episodes of hospital-treated self-harm by Indigenous status and one of a few that has compared outcomes between suicidal ideation and self-harm. The results point to several important implications for better understanding opportunities for preventing self-harm. Firstly, the lower absolute risk of repeat hospital admissions involving self-harm estimated in this study compared to others requires some clarification. The 1-year incidence of repeat hospital admissions involving self-harm in the NT we observed (8.9% and 9.4% amongst Aboriginal and non-Aboriginal people, respectively) is just over half of what was reported in a recent meta-analysis (17.0%) [22] and is the lowest reported in other hospital-based studies from Australia [23–25]. Study characteristics may explain this difference since studies that use hospital admissions data to define cohorts and outcome measures have tended to report a lower incidence of repeat episodes of self-harm [3]. This could be because the use of hospital admissions alone may be prone to underestimating the true incidence of hospital-treated self-harm due to the higher threshold of medical seriousness requiring admission [26,27]. Another factor could be the inclusion of suicidal ideation at inception, which is not common in other studies and which the results in this study show is associated with a lower risk of the outcome compared to self-harm at inception. However, of the methods associated with repetition from multivariable modelling, the estimated 1-year absolute risk of repetition for all self-harm amongst Aboriginal people (10.4%) and self-poisoning amongst non-Aboriginal people (11.1%) is still lower in this study compared to other studies. The relatively lower rate of repeat self-harm taken together with increasing trends in hospital admission involving suicidal ideation and self-harm in the NT observed during this study period [13] point to a potentially concerning increase in incident episodes that may suggest the need to also consider early intervention strategies to help prevent self-harm and subsequent suicide. However, it also possible that the results either point to potentially protective follow-up responses to hospital admissions involving suicidal ideation and self-harm or are an artefact of alternatives to hospital responses to repeat self-harm in community settings or in other parts of the healthcare system.

It should be noted that the absolute risk of repeat hospital admissions involving self-harm in the NT continued to increase over time compared to other studies, growing to around three quarters of the 5-year pooled incidence of repeat self-harm estimated by a meta-analysis of hospital-based studies (18.8% Aboriginal and 15.5% non-Aboriginal vs. 22.4% pooled estimated incidence) [3]. Whilst it has been suggested that the risk of repeat self-harm tends to plateau after 5 years [28], this study has shown continual long-term increases to almost 30% and just over 20% for Aboriginal and non-Aboriginal people, respectively. This suggests that the underlying risk of self-harm repetition observed in NT hospitals is likely the result of persistent or recurring long-term influences. In the absence of any association between enduring demographic characteristics that could explain these results, it is important that the broader context and longer-term influence of clinical characteristics are considered in drawing out implications for hospital management and further care in the community.

Substance misuse, especially involving alcohol, was especially prevalent and found to be associated with an increased risk of repeat hospital admissions involving self-harm. These results are consistent with existing evidence demonstrating a higher risk of a range of suicidal behaviours associated with alcohol use disorders [29]. Furthermore, the results appear to confirm both the proximal and distal influence of harmful alcohol consumption on the repetition of self-harm. That is, the association with previous admissions involving alcohol use disorders (for both Aboriginal and non-Aboriginal people) potentially points to the role of chronic misuse creating an underlying vulnerability to suicidal behaviour and the association with alcohol use disorders at index admission (non-Aboriginal people only) suggests that acute alcohol intoxication is a potential facilitator of suicidal behaviour and barrier to adopting alternative coping behaviours [30]. In light of the high rates of harmful alcohol consumption in the NT [31], the findings from this study reinforce the need for evidence-based population level policies to reduce harmful alcohol consumption that should contribute to reducing the risk of self-harm [32] and its repetition. Moreover, the very high burden of alcohol misuse on the hospital system in the NT [33] and the particularly high risk of self-harm associated with high levels of acute alcohol intoxication [34] suggests self-harm hospital admissions involving alcohol represent an important opportunity for referral to psychological interventions targeting misuse that are known to reduce the risk of subsequent self-harm in individuals [35]. Whilst further investigation is needed to determine the appropriateness and effectiveness of existing treatment services in the NT [36], it is even more important that mental health staff in the hospital are adequately supported to deal with the challenges posed by alcohol intoxication that often lead to poorer quality of care and non-assessment of self-harm [37].

The findings relating to mental health conditions diagnosed in non-Aboriginal people in the study point to important clinical risk factors that are known to increase vulnerability to self-harm [38] and help to clarify the potential opportunities for intervention. Psychological interventions, such as CBT, recommended for reducing vulnerability to self-harm repetition [39] should be considered as a therapeutic option as they have been shown to also effectively treat common mental disorders in self-harm, such as depression [40] and anxiety [41]. Furthermore, these treatments have also demonstrated effectiveness in reducing rates of self-harm repetition in the presence of personality disorder [42], especially borderline personality disorders, which this study suggests should also be a target for prevention in the NT. Since the NT has the highest rates of specialist mental health service use of any jurisdiction in Australia [43], further investigation into the quality and appropriateness of such care following hospitalisation for self-harm and suicidal ideation is warranted.

The association observed in non-Aboriginal people between repeat self-harm and severe mental disorders, comprising mostly schizophrenia and schizoaffective disorders, may require different approaches to clinical management depending on the stage and severity of psychosis or psychotic symptoms. The risk of self-harm is known to be greatest during early onset psychosis [44] and higher amongst those with other common risk factors for suicidal behaviour [45]. Furthermore, the existing evidence suggests that a higher risk of self-harm is associated with inadequate care and management in the community such as: poor adherence to medication [46], longer period of untreated psychosis [47,48], and higher frequency of psychiatric admissions [45]. These concerns may be particularly important for the NT where the use of depot antipsychotic medication for the treatment of psychotic illness has historically been the highest of any Australian jurisdiction [49]. The high use of depot medication may, among other things, reflect the application of clinical recommendations associated with widespread concerns with suicide risk in the NT clinical population [50]. For these reasons, comprehensive and integrated models of care, especially for people with early onset psychotic illnesses, are required that are known to assist with the management of significant mental health issues and reduce the risk of further suicidal behaviour [39].

The results from the stratified analysis point to distinct risks for repetition within Aboriginal and non-Aboriginal populations hospitalised for self-harm and suicidal ideation. The methods of self-harm associated with a higher risk of repetition differed by Indigenous status. Consistent with the evidence pointing towards higher risk of repetition associated with less fatal methods [51–53], self-poisoning was associated with a higher risk and hanging associated with a lower risk of repetition amongst non-Aboriginal people. For Aboriginal people, however, all methods of self-harm were associated with an increased risk of repetition. Since any form of self-harm repetition has been found to be associated with a higher risk of suicide [7], the findings from this study may point to overlaps between those at risk of repeat self-harm and suicide. Therefore, means restriction counselling to reduce the risk of repeat self-harm should be informed by comprehensive psychosocial assessment to ensure the effectiveness of such strategies are based on an understanding of the psychological and interpersonal contexts of intent and unmet needs. Younger age was associated with a higher risk of repetition for non-Aboriginal but not Aboriginal people, despite their younger age profile at index admission. Whilst the results for non-Aboriginal people are consistent with the evidence pointing to a higher frequency of repetition associated with the earlier onset of self-harm, especially in adolescence [54], issues of confounding or universally experienced influences may be at play for Aboriginal people. For example, it may be the case that clinically relevant associations, such as substance misuse and type of self-harm, may better explain age related differences in the results for Aboriginal people. Alternatively, it may be that Aboriginal people of all ages are universally affected by contextual influences not measured by this study, such as conditions of relative disadvantage, marginalisation and discrimination, cultural loss and intergenerational trauma [55]. However, the prevalence and role of alcohol misuse in this study more likely suggests the interplay of individual psychosocial stressors and underlying social determinants [56]. That is, the results from this study, taken together with Aboriginal perspectives, point to a combination of both individually experienced and collectively shared factors. Clinical management and aftercare for Aboriginal people should be informed by careful consideration of these influences, which requires culturally responsive assessment that balances both patient and cultural safety [57]. More importantly, investments in more culturally appropriate and safe services to effectively treat individuals in the community needs to be coupled with greater investments in self-determined and strengths-based prevention efforts targeting locally relevant issues and cultural supports [10]. Overall, this study suggests that the risk of repetition amongst non-Aboriginal people to be more closely associated with clinical populations for whom there is a need to ensure proper access to evidence-based treatments and effective services. But, for Aboriginal people, the higher risk of repeat hospital admission involving self-harm may be seen as the individual expression of adverse contextual influences requiring access to evidence-based individual treatments complemented by community-led cultural solutions.

Conclusion

This study highlights important similarities and differences in long-term risk of repeat hospital admission involving self-harm for Aboriginal compared to non-Aboriginal people in the NT. The role of alcohol and substance misuse for both Aboriginal and non-Aboriginal people may reflect a complex intersection of individual vulnerability and specific psychosocial stressors that requires further investigation to better determine the role that hospitals may play in encouraging individual treatment and promoting prevention at a population level. For non-Aboriginal people the higher risk of repetition also appears to be associated with clinical populations, pointing to the need to better understand the availability of evidence-based interventions and the effectiveness of existing services in reducing risk and supporting recovery. There is also an urgent need to ensure that Aboriginal people can access culturally safe and responsive options for effectively managing and treating the immediate unmet needs of individuals after discharge. This should be complemented by greater investments in community-based interventions that promote culture, resilience, and self-determination that may better address the long-term underlying risk for self-harm. These findings highlight the challenges for prevention in the NT and reinforces the importance for comprehensive psychosocial assessment in hospitals to identify risk, needs, and strengths distinct to Aboriginal and non-Aboriginal people to better inform appropriate and effective options for care both within and beyond the hospital setting.

Conflict of interest

All authors declare that they have no conflicts of interest.

Contributors

BL: Conceptualisation, methodology, formal analysis, data curation, writing – original draft, writing – review and editing, visualisation, project administration, funding acquisition; RB: Writing – review and editing; TH: Writing – review and editing; SS: Conceptualisation, methodology, writing – review and editing, funding acquisition; SG: Methodology, writing – review and editing; GR: Conceptualisation, methodology, writing – review and editing, funding acquisition.

Funding

BL was provided a research scholarship from the National Suicide Prevention Research Fund administered by Suicide Prevention Australia to undertake this work.

Supporting information

S1 Table. Descriptive summary and univariate analysis of repeat hospital admission involving self-harm by Aboriginal people with a hospital admission involving self-harm and suicidal ideation, NT 2001-2013

S2 Table. Descriptive summary and univariate analysis of repeat hospital admission involving self-harm by non-Aboriginal people with a hospital admission involving self-harm and suicidal ideation