Modeling Naturalistic Clusters of Adolescent Substance Abuse Treatment Services: A Latent Class Analysis

• Activity Number: 584
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Megan Suzanne Schuler*+ and Elizabeth Stuart and Beth Ann Griffin and Rajeev Ramchand
• Companies: Johns Hopkins Bloomberg School of Public Health and Johns Hopkins Bloomberg School of Public Health and RAND Corporation and RAND Corporation
• Keywords: latent class analysis ; adolescent ; substance use ; substance use treatment

Abstract

There is considerable variability in the services youth enrolled in drug treatment programs receive, due to variability in client needs, services offered or client compliance. Modeling treatment as a latent variable more fully captures constellations of treatment services received by adolescents. Data are from a multisite longitudinal observational study of 17,490 youth enrolled in drug treatment programs. Youth are assessed at study visits with the Global Assessment of Individual Needs (GAIN). 30 GAIN items relating to substance abuse treatment services were used as class indicators. Latent class regression was performed to identify characteristics associated with latent class membership. Analysis was conducted in Mplus. A 4-class model provided the best fit; classes were categorized as 1) any residential treatment, high service utilization, 2) only outpatient services, high services 3) only outpatient services, low service utilization and 4) no treatment services. Estimated class prevalences are 15%, 23%, 37% and 25% respectively, indicating that the majority of youth in this dataset were receiving only outpatient services, which is consistent with national trends.