Evaluation of a screening algorithm using the Strengths and Difficulties Questionnaire to identify children with mental health problems: A five-year register-based follow-up on school performance and healthcare use

Background: Treatment of mental health problems (MHP) is often delayed or absent due to the lack of systematic detection and early intervention. This study evaluates the potential of a new screening algorithm to identify children with MHP. Methods: The study population comprises 2,015 children from the Copenhagen Child Cohort 2000 whose mental health was assessed at age 11–12 years and who had no prior use of specialised mental health services. A new algorithm based on the Strengths and Difficulties Questionnaire (SDQ) is utilised to identify MHP by combining parent-reported scores of emotional and behavioural problems and functional impairments. The screening is done on historical data, implying that neither parents, teachers nor health care professionals received any feedback on the screening status. The screening status and results of an IQ-test were linked to individual-level data from national registries. These national registers include records of each child’s school performance at the end of compulsory schooling, their health care utilisation, as well as their parents’ socio-economic status and health care utilisation. Results: 10% of the children screen positive for MHP. The children with MHP achieve a significantly lower Grade Point Average on their exams, independently of their IQ-score, perinatal factors and parental characteristics. On average, the children with MHP also carry higher health care costs over a five-year follow-up period. The higher health care costs are only attributed to 23% of these children, while the remaining children with MHP also show poorer school performance but receive no additional health care. Conclusions: The results demonstrate that children with MHP and a poor prognosis can be identified by the use of the brief standardised questionnaire SDQ combined with a screening algorithm.