Mortality in children and adolescents following presentation to hospital after non-fatal self-harm in the Multicentre Study of Self-harm: a prospective observational cohort study

Abstract

Background: Self-harm and suicide in children and adolescents are growing problems, and self-harm is associated with a significant risk of subsequent death, particularly suicide. Long-term follow-up studies are necessary to examine the extent and nature of this association.

Methods: For this prospective observational cohort study, we used data from the Multicentre Study of Self-harm in England for all individuals aged 10-18 years who presented to the emergency department of five study hospitals in Oxford, Manchester, and Derby after non-fatal self-harm between Jan 1, 2000, and Dec 31, 2013. Deaths were identified through the Office for National Statistics via linkage with data from NHS Digital up until Dec 31, 2015. The key outcomes were mortality after presentation to hospital for self-harm, categorised into suicide, accidental deaths, and death by other causes. We calculated incidence of suicide since first hospital presentation for self-harm and used Cox proportional hazard models to estimate the associations between risk factors (sex, age, previous self-harm) and suicide.

Findings: Between Jan 1, 2000, and Dec 31, 2013, 9303 individuals aged 10-18 years presented to the study hospitals. 130 individuals were excluded because they could not be traced on the national mortality register or had missing data on sex or age, thus the resulting study sample consisted of 9173 individuals who had 13 175 presentations for self-harm. By the end of the follow-up on Dec 31, 2015, 124 (1%) of 9173 individuals had died. 55 (44%) of 124 deaths were suicides, 27 (22%) accidental, and 42 (34%) due to other causes. Of the 9173 individuals who presented for self-harm, 55 (0·6%) died by suicide. Most suicide deaths involved self-injury (45 [82%] of 55 deaths). Switching of method between self-harm and suicide was common, especially from self-poisoning to hanging or asphyxiation. The 12-month incidence of suicide in this cohort was more than 30 times higher than the expected rate in the general population of individuals aged 10-18 years in England (standardised mortality ratio 31·0, 95% CI 15·5-61·9). 42 (76%) of 55 suicides occurred after age 18 years and the annual incidence remained similar during more than 10 years of follow-up. Increased suicide risk was associated with male sex (adjusted hazard ratio 2·50, 95% CI 1·46-4·26), being an older adolescent at presentation to hospital for self-harm (1·82, 0·93-3·54), use of self-injury for self-harm (2·11, 1·17-3·81; especially hanging or asphyxiation [4·90, 1·47-16·39]), and repeated self-harm (1·87, 1·10-3·20). Accidental poisoning deaths were especially frequent among males compared with females (odds ratio 6·81, 95% CI 2·09-22·15).

Interpretation: Children and adolescents who self-harm have a considerable risk of future suicide, especially males, older adolescents, and those who repeated self-harm. Risk might persist over several years. Switching of method from self-harm to suicide was common, usually from self-poisoning to self-injury (especially hanging or asphyxiation). Self-harm is also associated with risk of death from accidental poisoning, particularly involving drugs of abuse, especially in young males.

Funding: UK Department of Health and Social Care.