Risk Score for Predicting Adolescent Mental Health Problems Among Children Using Parental Report Only: The TRAILS Study

Abstract

ObjectiveTo construct a risk score for adolescent mental health problems among children, using parental data only and without potentially stigmatizing mental health items. MethodsWe prospectively derived a prediction model for mental health problems at age 16 using data from parent report on 1676 children aged 11 from the general population. Mental health problems were considered present in the top 15% scores on the combined Achenbach ratings. The model was validated in a separate cohort (n = 336) children. A risk score was constructed for practical application. ResultsIn the derivation cohort, 248 (14.8%) had mental health problems at follow-up. Predictors in the final model were gender, maternal educational level, family history of psychopathology, math achievement at school, frequently moving house, severe disease or death in the family, parental divorce, and child frustration level. The model was well calibrated, showed good discriminatory power (area under the curve 0.75; 95% confidence interval 0.72–0.78), and validated well. The risk score stratified children in classes of risk ranging from 6.6% to 52.2%. ConclusionsA risk score based on parent-reported data only and without mental health items accurately estimated the 5-year risk of adolescent mental health problems among children from the general population. Children with high risk may benefit from further monitoring or intervention. The risk score may be particularly suitable when parents want to circumvent an explicit discussion on possible mental health problems of their child.