School absence and mental health service referral: Cohort study of a South London data linkage

Background The longitudinal relationship between school absence and mental health has important policy implications; if school absence predicts later mental health problems, it could be used to identify young people at increased risk and enable further assessment, prevention, and early intervention. Methods We analysed an existing data linkage between the National Pupil Database and healthcare records representing a sample of 47,926 young people aged 11–15 in the UK. We used logistic regression to examine the longitudinal association between persistent school absence (defined by the Government Department for Education in England as missing more than 10% of available school sessions) and later contact with secondary care mental health services. We also compared the sensitivity and positive predictive value of the >10% absence threshold to alternative thresholds for predicting adverse mental health outcomes. Results At the currently applied threshold of >10%, persistent school absence was associated with 2.77 (95% CI 2.33–3.30, girls) and 1.58 (95% CI 1.29–1.95, boys) times the odds of accepted referral to secondary care mental health services in the following year, after adjustment for sociodemographic and educational factors. The absolute risk difference for girls was 4.7% (95% CI 3.9–5.4) and for boys, 2.8% (95% CI 2.1–3.4). Compared to other thresholds, the >10% absence threshold provided a good balance between sensitivity (40.5%), positive predictive value (5.5%), and identifying a manageable proportion of young people as high risk (18%). Conclusion The longitudinal relationship between school absence and later secondary care mental health service contact suggests that school absence may be a useful marker for educators to identify children in need of support. The >10% absence threshold used in education policy in England may also serve as a useful marker of later mental health risk in other national policy contexts.