Helping Policymakers Better Predict the Prevalence of Children's Mental Health Conditions: An Analysis Using Restricted and Publicly Available Health and Education Data

Music City Center 

One in five children has a mental health condition, yet less than half of these children receive needed treatments and services. Many state and local policymakers need tools to better identify and prioritize children most at risk of mental health conditions and their negative consequences.  We use publicly available, county- and school district-level datasets along with linked, health and education administrative from the State of Tennessee to predict the prevalence of mental health conditions among Tennessee children. These data include diagnoses of mental health conditions at the child level, allowing us to assess the prevalence of these conditions (by child age) across the state. Recognizing, however, that many policymakers making decisions about the distribution of finite resources do not have access to these rich, linked data systems, we address the question: in the absence of individual-level administrative data on child mental health, can policymakers predict where children's mental health needs are greater?  We separately predict the prevalence of mental health conditions among children first using only non-publicly available data and then using only publicly available county-level measures. Next, we combine measures from both sets of data to perform the same analysis and compare the predictive models' performance across these specifications. We assess whether there are key variables that consistently predict the prevalence of children's mental health needs/conditions across the estimations and whether there are publicly available data that could provide a sufficient guide for the allocation of resources to better meet children's mental health needs. 

Children

Mental health