Abstract:
|
Project INTEGRATE is an ongoing research synthesis project that leverages existing data from individual-focused brief alcohol interventions to draw robust inferences about which interventions work well for young adults and how they work. We have extracted study-level aggregate data from 124 studies (2013 - 2018) and item-level data for participants from 24 heterogeneous studies (12,630 participants). We have incorporated treatment arms, covariates, outcomes, and follow-ups within studies in analysis to provide more precise answers and utilized various methods, including two-parameter logistic item response theory modeling for developing commensurate measures, Bayesian multilevel modeling, multivariate random-effects meta-analysis via confidence distribution-based mapping, and complex survey structural equation modeling. Based on results thus far, individually delivered motivational interviewing interventions that incorporate personalized feedback are better than group-based or feedback-only interventions for reducing alcohol-related problems and driving after heavy drinking. Data and R codes have been shared to promote new or alternative statistical methods. Support: R01 AA019511.
|