Strengths Model for Youth: Initial Longitudinal Analysis on Youth Outcomes

Background/Purpose: Before 2014, no formal case management model had been implemented with the adolescent mental health consumer population in Kansas. In 2014, researchers began the process of piloting the Strengths Model for Youth (SM-Y), an adapted version of the adult Strengths Model, for work with adolescents 12 and older diagnosed with a mental illness. Since 2014, researchers have collected monthly adolescent outcomes pertaining to hospitalizations, socialization, and academic performance. This study presents initials findings on implementation of SM-Y and its relationship to those identified outcome areas. Specifically, this study uses multilevel modeling (MLM) to estimate change in probabilities for adolescents in hospitalization, socialization, and educational performance.

Methods: Beginning in 2014, researchers partnered with one community-based mental health center in Kansas. Utilizing a staggered implementation approach, researchers provided initial SM-Y training and began implementation of SM-Y with CMHC staff. By 2017, SM-Y was implemented across all five CMHC’s adolescent serving teams and was delivered to adolescents 12 and older. The number of adolescents served varied by each team due to variations in the number of direct service workers comprising the team. Adolescent outcomes were collected on a monthly basis. Due to the staggered nature of implementation, the number of outcomes collected prior to implementation and post implementation varied for each team. Therefore, for analysis purposes, only implementation months four through 36 are included. The number of adolescents included in the sample ranges from 711 to 772. Researchers estimated change in logits for hospitalization, socialization, and education using multilevel modeling (MLM) and the SAS Proc GLIMMIX procedure. Individual trajectories for logits and probabilities of socialization, education and hospitalization were modeled. First, single level empty means model was estimated, and followed by an empty means random intercept model. Then a fixed time, random intercept model was estimated. To track curvilinear relationship of time with outcome variables, quadratic and cubic time variables were added to the model. Team dummy variables were also included to capture between-team variances. To determine best fitting model, -2 Log Likelihood tests were implemented and to verify the fixed effect, multivariate Wald tests were used.

Results: In regard to hospitalization, results indicate after SM-Y implementation, the predicted logit of an adolescent not being hospitalized had a curvilinear pattern with time. This means that post SM-Y implementation, the probabilities of an adolescent not being hospitalized grows with time. For socialization and for education, models demonstrate that adolescents were more likely to have engaged in a social activity and have average or above average grades post-implementation of SM-Y.

Implications: The initial findings from this study are encouraging and have significant implications for future social work practice with adolescents in community mental health. Results demonstrate using SM-Y can have significant and positive effects in reducing the probability that an adolescent will experience intensive inpatient treatment while simultaneously experience an increase in socialization and academic achievement. While these results are preliminary and more analysis is needed, results indicate SM-Y is an effective practice for community mental health which is need of empirically supported practices.