Trauma, Substance Use, and Mental Health in the Child Welfare System: A Latent Transition Analysis

Adolescents in the child welfare system are at notably elevated risk for substance use compared to adolescents in the general population (Fettes, Aarons, & Green, 2013). Those who are placed into the system are often suspected to have had one or more potentially traumatic (and illegal) events inflicted upon them by their caretakers, such as neglect, sexual abuse, other physical abuse, or psychological abuse. Despite the considerable implications of trauma for substance use, preventive efforts often focus myopically on substance use alone rather than its critical comorbidities.

This study sought to document developmental patterns of substance use, mental health, and trauma in a representative national sample of adolescents in the child welfare system. A subsample of adolescents initially 11-14 years old (n = 1,178) was taken from the National Survey of Child and Adolescent Well-Being (NSCAW). Three-step latent mixture models (Asparouhov & Muthén, 2014) were computed from data collected at each of three time points within three years of the close of the adolescent's case investigation. Indicators included self-reported past year misuse of cigarettes, alcohol, marijuana, and prescription painkillers, and t-scores from the Children's Depression Inventory (Kovacs, 1992), Trauma Symptom Checklist for Children (Briere, 1996), and the Child Behavior Checklist (child and parent versions; Achenbach, 1991). Class covariates included gender, age, race, type(s) of maltreatment experienced in the past year, and the total number of types of maltreatment. A latent transition analysis was subsequently computed to model the transition probabilities between classes at different time points, providing a better picture of the quality of substance use and mental health development over time.

Results from the mixture models yielded five classes at the first time point, four at the second, and five at the third. Adolescents who had the highest rates of depression, trauma, and internalizing/externalizing symptoms had only moderately high probabilities of depression. Similar patterns were observed at the subsequent two time points. The LTA revealed considerable variability in transition probabilities. A few transitions were especially common, including: 1) movement from low issues at the first two time points to much greater severity by the third time point, 2) movement from low issues at the first two time points to slightly worse externalizing symptoms by the third, and 3) consistently high severity across time points. Sexual maltreatment was associated with the most severe outcomes at each time point, though numeracy of types of maltreatment experienced was also strongly associated with more negative outcomes. Additional patterns of transitions and covariates are discussed.