(PS6-18) Latent Classes of Practice Elements Utilized for Behavioral Health Treatment of High School Students with Disruptive Behavior

Disruptive behavior problems among adolescents, including anger, aggression, oppositionality, and defiance, are a common target of mental health treatment with a wide range of negative sequalae (Steiner et al., 2017). Additionally, previous literature has found multi-domain interventions that target a variety of settings in which such youths interact to be the most efficacious approaches for these types of difficulties among adolescents (McCart, Sheidow, & Jaramillo, 2023). As such, understanding the types of interventions that youth with disruptive behavior problems are receiving through school-based systems of care is essential for efforts aimed at targeting these difficulties. Hawaii’s Department of Education (DOE) School-Based Behavioral Health (SBBH) system recently implemented an electronic health record system in which therapists record intervention details, including the specific practice elements used. The primary aim of this study is to identify patterns of practice elements that youth with disruptive behavior problems are receiving through school-based supports, to develop a greater understanding of the types of interventions that these youth are currently receiving in the school setting.

Clinical data for 726 high school students who received SBBH services in Hawaii’s DOE system from January 2023 to January 2024 were analyzed. Latent class analyses were conducted to group youth into relatively homogenous classes based on the specific practice elements that were used as part of their SBBH services. The ten most commonly endorsed practice elements for this sample were selected as indicators for all latent class models (i.e., Communication Skills, Problem Solving, Goal Setting, Cognitive/Coping, Natural and Logical Consequences, Relationship/Rapport Building, Therapist Praise/Rewards, Self-Monitoring, Mindfulness, and Supportive Listening). To select the model that best explained the data, multiple latent class analyses were conducted. An initial one-class unconditional model was computed first, with the number of classes successively increased until the models no longer converged. Relative model fit for each model was assessed using a variety of model fit indices (e.g., Bayesian information criterion). Additionally, classification diagnostics (e.g., entropy) and theoretical implications of competing models were considered, before selection of a final model.

 


            A four-class model was selected as best fitting the data, due to superior fit indices compared to the three-class model (e.g., higher entropy, BLRT-test  (p < .05 compared to three-class model); although the five-class model was also considered, the four-class model was ultimately selected due to the five-class model containing one class with a relatively small size (i.e., n > 40). The four classes were identified as: 1) General Supports (24.1%), 2) Individual Coping Skills (16.1%), 3) Supportive Listening and Rapport Building (19.8%), and 4) Other Practices (39.9%). Descriptions of the four classes will be discussed, along with implications for school-based disruptive behavior interventions for high school aged youth.