Risk factors for nonfatal self-harm and suicide among adolescents: two nested case–control studies conducted in the UK Clinical Practice Research Datalink

Abstract

Background: The characteristics of adolescents who die by suicide have hitherto been examined in uncontrolled study designs, thereby precluding examination of risk factors. The degree to which antecedents of nonfatal self-harm and suicide at young age differ remains unknown. Method: We delineated two nested case–control studies of patients aged 10–19 years using the Clinical Practice Research Datalink with interlinked hospital and national mortality records. Cases were adolescents who between 1st January 2003 and 31st December 2018 had died from suicide (N = 324) – study 1; experienced their first self-harm episode (N = 56,008) – study 2. In both studies, cases were matched on sex, age and practice-level deprivation quintile to 25 controls. By fitting conditional logistic regression, we examined how risks varied according to psychiatric diagnoses, prescribed psychotropic medication, patterns of clinical contact and area-level deprivation. Results: Suicides occurred more often among boys (66%), but self-harm was more common in girls (68%). Most individuals who self-harmed or died from suicide presented to their GP at least once in the preceding year (85% and 75% respectively). Only a third of cases had one of the examined diagnostic categories recorded. Depression was most strongly associated with elevated risks for both outcomes (self-harm: OR 7.9; 95% CI 7.8–8.2; suicide: OR 7.4; 95% CI 5.5–9.9). Except for autism spectrum disorder, all other diagnostic categories were linked with similar risk elevations for self-harm as for suicide. Whilst self-harm risk rose incrementally with increasing levels of area-level deprivation, suicide risks did not. Conclusions: We observed few marked differences in risk factor profiles for nonfatal self-harm versus suicide. As most adolescents who had harmed themselves or died by suicide were known to services in the preceding year, their underlying pathology may not be adequately identified and treated. Our findings highlight the need for a multiagency approach to treatment and prevention.