Abstract
BACKGROUND: Sleep problems and self-harm during adolescence are both highly common and major public health concerns, yet the nature of their relationship remains poorly understood. This study examined the cross-sectional and long-term effects of several sleep phenotypes on self-harm and whether decision-making moderated this relationship. METHODS: Data was utilised from the Millennium Cohort Study (n = 10,477, Female = 5,314 [50.72%]) when individuals were approximately 9 months, 14 years and 17 years of age. Sleep variables available were measured at 14 years and included sleep duration on weekdays and weekends, social jetlag, sleep onset latency and night awakenings. Self-harm was measured at 14 and 17 years. The Cambridge Gambling Task assessed rational decision-making at 14 years. RESULTS: Cross-sectionally, shorter sleep duration on school days (AOR = 0.875; 95% CI = 0.820, 0.933; p < .001), longer sleep onset latency (AOR = 1.005; 95% CI = 1.002, 1.007; p < .001) and more frequent night awakenings (AOR = 1.140; 95% CI = 1.086, 1.197; p < .001) were significantly associated with self-harm at 14, even when controlling for demographic and clinical covariates. Longitudinal results indicated that shorter sleep duration on school days (AOR = 0.926; 95% CI = 0.874-0.982; p = .010), longer sleep onset latency (AOR = 1.003; 95% CI = 1.001-1.005; p = .008) and more frequent night awakenings (AOR = 1.090; 95% CI = 1.043-1.139; p < .001) also had a direct prospective effect on self-harm at 17 when controlling for demographic and clinical factors. Rational decision-making as measured by the Cambridge Gambling Task did not significantly contribute to this relationship. CONCLUSIONS: These findings highlight the prospective association between short sleep duration, increased sleep onset latency, fragmented sleep and self-harm during adolescence. Ensuring adolescents obtain enough good quality, uninterrupted sleep appears critical to prevent engagement with self-harm.