Summary of the findings

In this study, we undertook an analysis of self-harm-related mental health presentations to emergency departments by children and young people (CYP) aged 10–17 years, in six public hospitals in the South Western Sydney Local Health District (SWSLHD) from 2016 to 2022. We found a steady increase in the proportion of self-harm-related mental health presentations to emergency departments over the 6-year study period, from 27 to 39%. CALD background was associated with lower odds of presenting to the mental health emergency department for self-harm, after adjusting for potential covariates and individual-level clustering. There were also significant associations between self-harm-related mental health emergency department presentations and older age, female gender, weekday presentations and the COVID-19 period.

Comparison with existing literature

Almost one in three (31.5%) of all mental health-associated emergency department presentations by CYP were related to self-harm, alongside a steady increase during the study period. The increasing number of self-harm presentations to the emergency department by CYP observed in this study is consistent with the recent literature.^{Reference Sara, Wu, Uesi, Jong, Perkes and Knight16,Reference Hiscock, Chu, O'Reilly, Freed, White and Danchin24} This worsened trend might be attributed to the effects of the COVID-19 pandemic, as our study, consistent with local^{Reference Sara, Wu, Uesi, Jong, Perkes and Knight16,Reference Hiscock, Chu, O'Reilly, Freed, White and Danchin24} and international evidence,^{Reference Madigan, Korczak, Vaillancourt, Racine, Hopkins and Pador25} found that the pandemic period was significantly associated with higher odds of presentation. These exacerbated rates of self-harm presentation among CYP might be due to school closures, lack of social gatherings, family stress and domestic violence.^{Reference Sedgwick, Epstein, Dutta and Ougrin6}

Findings of this study particularly looked at the association of self-harm-related mental health emergency department presentations among CYP from culturally and linguistically diverse (CALD) backgrounds. We found that CYP from a CALD background – either CALD status (country of birth and/or preferred language) or preferred language only (non-English) – had lower odds of presenting to the emergency department for self-harm compared with those from a non-CALD English-language background. However, CALD status based on country of birth was not a significant factor. This may be due to the fact that many of the children born in Australia have families from CALD backgrounds. Evidence on CALD status and self-harm remains unclear, with some studies reporting higher rates of presentation, particularly among refugees and asylum seekers.^{Reference Pham, O'Brien, Berecki-Gisolf, Liu, Gibson and Clapperton13,Reference Hedrick and Borschmann26} However, other studies have shown reduced presentations among CALD groups owing stigma around mental illness within families and the community,^{Reference Chang, Steeg, Kapur, Webb, Yip and Cooper27} concerns about care provision and language barriers.^{Reference Gater, Waheed, Husain, Tomenson, Aseem and Creed28} Additionally, CALD CYP's experiences of discrimination or racism and their struggle with a sense of belonging can contribute to risk of self-harm and health service avoidance.^{Reference Sawrikar and Hunt29} Further, other studies report that CALD CYP's social networks and supports, religious beliefs and individual resilience could be some of the protective factors against self-harm.^{Reference Cooper, Enticott, Shawyer and Meadows30,Reference Khatri and Assefa31} Nevertheless, further investigation is needed to determine whether the low rates of presentation from CALD backgrounds reflect actual lower rates, or are due to stigma and/or differences in help-seeking behaviours. Additionally, research to better understand the variation by taking into account key factors such as the mode of entry into Australia (refugee versus planned migration), complemented by detailed qualitative interviews of CALD CYP, is critical. This will deepen the understanding about how different self-harm prevention strategies are needed for different CALD populations and aid in better supporting the specific needs of CYP from CALD communities most affected by self-harm.

Consistent with the findings of other studies,^{Reference Hiscock, Chu, O'Reilly, Freed, White and Danchin24,Reference Poyraz Fındık, Erdoğdu, Fadıloğlu and Rodopman Arman32} we found key sociodemographic factors such as older age and female gender to be significant predictors of self-harm-related emergency department presentations. The predominance of female adolescents with increased presentations has been well established, with a previous study reporting 6:1 and 3.5:1 female to male ratios at 12–14 years and 15–17 years respectively.^{Reference Diggins, Kelley, Cottrell, House and Owens33} This is also reflected in our descriptive findings, where the numbers of female adolescents presenting for overall mental health problems and self-harm were double those of males. Although it could be true that female adolescents could have higher mental health service use than male adolescents, this could also be due to higher rates of help-seeking behaviours and readiness to discuss emotional problems among female adolescents compared with males.^{Reference Liu, Jia, Qin, Zhang, Yu and Luo34} However, other risk factors for this gender inequality may also include higher reports of poor self-image, bullying at school and via social media, academic pressure, abuse, and neglect at home.^{Reference Hartas35} Hence, it is essential to continue considering gender-related differences and characteristics to better understand self-harm behaviours in the adolescent population.

Interestingly, we also found that the weekdays (compared with weekends) were significantly associated with higher odds of self-harm-related mental health emergency department presentations among CYP. Only a few studies have reported on the timing of self-harm-related emergency department presentations among CYP, of which Sara et al's study^{Reference Sara, Wu, Uesi, Jong, Perkes and Knight16} mentioned that presentations peaked on Mondays and declined during the week to a low on Saturdays. In line with this, Colman et al's study^{Reference Colman, Yiannakoulias, Schopflocher, Svenson, Rosychuk and Rowe36} also reported that Sundays and Mondays were the most common days for mental health presentations. This may partially be attributable to fewer out-patient services being available during the weekend. It may be speculated that this may be also due to CYP being more likely to encounter an adverse event relating to friendship or relationship issues, as well as academic stresses, which might trigger self-harm behaviours while at school on weekdays, compared with the weekends when they are with families. Furthermore, on weekends, CYP may spend more time with their families, which can serve as a protective factor against self-harm-related behaviours by providing family support.^{Reference Fortune, Cottrell and Fife37}

Methodological reflections

Our study has several strengths. The data are drawn from a large clinical and administrative data-set from six public hospitals in SWSLHD, an area of significant social and economic disadvantage and high proportion of CALD population groups. This study used an encounter-level analysis as opposed to a patient-level analysis, making it possible to evaluate encounter-level factors in relation to self-harm-related presentations. The findings of this study emphasise the importance of enhancing culturally appropriate mental health services in the community, to alleviate the burden associated with self-harm among adolescents in SWSLHD. We also note several limitations to our study. First, our study is based on data from six public hospitals, hence we cannot comment on encounters in the community or emergency department presentations to other hospitals outside the local health district. Second, the findings demonstrate an association rather than a causative relationship between predictors and outcomes. Third, we defined the CALD status using the Australian Bureau of Statistics’ minimum core set of cultural and language indicators, such as preferred language and/or country of birth. However, we acknowledge that there are other identifiers for CALD status, for instance the reason for migration (e.g. skilled migration versus refugee/asylum seekers). Hence, the decreased odds of self-harm presentation among the CALD communities may not be a true representation of different social and ethnic groups. Further, the association between CALD status and self-harm might be attributable to unmeasured confounders, and hence their relationship needs to be interpreted with caution. Finally, unlike a study^{Reference Cwik, Barlow, Tingey, Larzelere-Hinton, Goklish and Walkup38} which explored the type of self-harm (suicidal attempts versus non-suicidal self-injuries), this information along information on repeat presentations was not available for analysis, given the nature of the administrative data. Hence, there is opportunity for future research to further explore in detail how people of different CALD backgrounds seek health services for different mental health issues and the mental health service needs they have.

Implications of the findings

Our study found a year-on-year increase in the proportion of self-harm-related emergency department presentations by CYP in SWSLHD. Although our primary analysis showed lower odds of self-harm-related emergency department presentations by CYP from a CALD background, other risk factors, such as older age, female gender and the pandemic period, were associated with higher odds of presentation. Nevertheless, these findings demonstrate that if service provision has remained static over this time, then the increasing demand may be challenging the capacity of services.^{Reference Khan, John, Middleton, Huang, Lin and Hu22} Further, Geulayov et al^{Reference Geulayov, Casey, McDonald, Foster, Pritchard and Wells39} reported on the iceberg model of self-harm, where among CYP in England the rate of community-reported self-harm was 10 times higher than the rate of hospital-presenting self-harm. Thus, the true extent of the overall prevalence of self-harm is likely to be much higher in the community than that reflected in the emergency department presentations reported here. In line with this, the recent establishment in Australia of the National Mental Health Commission's National Suicide and Self-Harm Monitoring System may assist in defining/measuring self-harm presentations in the future.⁴⁰ With individuals and families finding it increasingly difficult to access and navigate suitable services (hence ending up in the emergency department for acute care), there is an urgent need to implement innovative service models that are responsive, integrated, equitable and sustainable.

Alternative service models to the traditional emergency department and hospital-based service delivery also deserve attention in this regard. In line with this, NSW Health's recent implementation of alternative models of care for emergency department avoidance, such as Towards Zero Suicide (TZS) and the ‘Safeguards’ child and adolescent mental health rapid response teams, may help redirect the patient load away from emergency departments. Further, targeted supports in a less formal non-emergency department community setting, including the involvement of people with lived experience as peer workers, may be preferred by CYP and such models would benefit from robust evaluation.