This K02 competitive continuation application proposes to advance understanding of adolescent alcohol use disorders (AUDs) through examination of adult outcomes with innovative statistical techniques for longitudinal data. In Years 6 through 10, the program focus evolves from statistical techniques for categorical variables to techniques based on finite mixture models. The programmatic focus shifts from relationships among adolescent characteristics to adult outcomes. The project will emphasize person-centered statistical strategies to predict adult outcomes among adolescents characterized by substance involvement trajectories, psychopathology latent classes, childhood maltreatment histories, and treatment participation. With NIAAA and NIDA support, the Pittsburgh Adolescent Alcohol Research Center (PAARC) is completing adult follow-up assessments in adolescents with AUDs and a reference group. The hypotheses consider the confluence of behavioral, cognitive, and affective dysregulation in predicting AUD course, complications, and outcomes. Adult outcomes examined include AUDs, other substance use disorders, psychopathology,) personality disorders, academic achievement, work performance, family status, somatic symptoms, liver injury, and sexually transmitted diseases. Complementary data from the Center for Education and Drug Abuse Research (NIDA P50: PI Tarter, Co-PI Clark), a study of children at risk for alcohol and drug use disorders followed through adolescence into adulthood, will also be utilized. Incorporating categorical variables pertinent to symptom and diagnostic categories, the statistical strategies applied will include multivariate survival analysis, general growth mixture models, and Bayesian instance-based machine learning. These methods will be specifically optimized to construct clinically relevant models applicable to individual patients, serving as the basis for more thorough phenotype description, more effective prevention programs, and individually tailored treatment interventions.
Showing the most recent 10 out of 72 publications