Predicting Service Urgency in Children and Youth with Autism Spectrum Disorder: The Development of an Algorithm

Author(s): Shannon L. Stewart, Gabrielle K.C. King, Jocelyn N. Van Dyke, Jeffrey W. Poss

Background: Autism spectrum disorder (ASD) is often accompanied by various mental health-related symptoms that impact a child’s level of functioning. This variability leads to unequal service urgency needs among children and youth with ASD. However, there is no empirically-based and clinically-informed system that assesses the urgency needs of individuals with ASD seeking mental health services. The current study describes the methods used to develop an algorithm to determine mental health service urgency for children and youth with ASD.

Method: Assessments from 20,781 and 53,387 children and youth, drawn from two types of the interRAI Child and Youth Mental Health (ChYMH) instruments used in Ontario, were examined to identify service urgency among those with a probable ASD diagnosis. An interactive decisiontree tool identified rules-based groups that were subsequently assigned to a manageable number of levels using k-means clustering. The fit of the algorithm was assessed by logistic regression analyses.

Results: The algorithm used twelve items to predict service urgency among 1,598 children with a probable ASD diagnosis. The decision tree identified 18 groups that were collapsed to five levels, from lowest to highest service urgency. The highest urgency group, which includes 12.1% of children and youth with probable autism and for whom 36.1% are rated urgent, is 11.6 times more likely to be rated as urgent, compared to the lowest group.

Conclusions: By implementing the first empirically and clinically supported decision-support tool, appropriate and efficient access to community-based mental health resources can be allocated to children and youth with ASD.