Predictors of future suicide attempt among adolescents with suicidal thoughts or non-suicidal self-harm: a population-based birth cohort study

Background: Suicidal thoughts and non-suicidal self-harm are common in adolescents and are strongly associated with suicide attempts. We aimed to identify predictors of future suicide attempts in these high-risk groups. Methods: Participants were from the Avon Longitudinal Study of Parents and Children, a population-based birth cohort study in the UK. The sample included 456 adolescents who reported suicidal thoughts and 569 who reported non-suicidal self-harm at 16 years of age. Logistic regression analyses were used to explore associations between a wide range of prospectively recorded risk factors and future suicide attempts, assessed at the age of 21 years. Findings: 38 (12%) of 310 participants with suicidal thoughts and 46 (12%) of 380 participants who had engaged in non-suicidal self-harm reported having attempted suicide for the first time by the follow-up at 21 years of age. Among participants with suicidal thoughts, the strongest predictors of transition to attempts were non-suicidal self-harm (odds ratio [OR] 2·78, 95% CI 1·35–5·74; p=0·0059), cannabis use (2·61, 1·11–6·14; p=0·029), other illicit drug use (2·47, 1·02–5·96; p=0·045), exposure to self-harm (family 2·03, 0·93–4·44, p=0·076; friend 1·85, 0·93–3·69, p=0·081), and higher levels of the personality type intellect/openness (1·62, 1·06–2·46; p=0·025). Among participants with non-suicidal self-harm at baseline, the strongest predictors were cannabis use (OR 2·14, 95% CI 1·04–4·41; p=0·038), other illicit drug use (2·17, 1·10–4·27; p=0·025), sleep problems (waking in the night 1·91, 0·95–3·84, p=0·069; insufficient sleep 1·97, 1·02–3·81, p=0·043), and lower levels of the personality type extraversion (0·71, 0·49–1·03; p=0·068). Interpretation: Most adolescents who think about suicide or engage in non-suicidal self-harm will not make an attempt on their life. Many commonly cited risk factors were not associated with transition to suicide attempt among these high-risk groups. Our findings suggest that asking about substance use, non-suicidal self-harm, sleep, personality traits, and exposure to self-harm could inform risk assessments, and might help clinicians to identify which adolescents are at greatest risk of attempting suicide in the future. Funding: American Foundation for Suicide Prevention, National Institute for Health Research Biomedical Research Centre at the University Hospitals Bristol National Health Service Foundation Trust, and the University of Bristol.