Decision-making skills as a mediator of the #Tamojunto school-based prevention program: Indirect effects for drug use and school violence of a cluster-randomized trial

Background: The aim of the present study was to evaluate a formal mediation analysis effect of the #Tamojunto program on adolescents' drug use and violent behavior in schools through decision-making skills using a potential outcomes approach. Methods: An in-cluster randomized controlled trial was conducted in 2014-2015 with 6691 7th- and 8th-grade students in 72 public schools in 6 Brazilian cities to evaluate the effects of the European drug prevention program Unplugged, called #Tamojunto in Brazil. Baseline data were collected prior to program implementation, and follow-up data were collected 9 and 21 months later. Mediation analysis using a potential outcomes approach, in which counterfactuals are modeled if positivity is met, was used to evaluate the indirect effects of the program #Tamojunto on the third-wave of drug use (alcohol, tobacco, marijuana, inhalants, and binge drinking) and school violence (bullying or physical, verbal and sexual aggression) assessment through decision-making skills. Results: When controlling for all covariates, the Total Natural Indirect Effect (TNIE) was significant only for pastyear drug use (TNIE = 0.003, 95%CI = 0.001; 0.007). In the adjusted models, 37.5% of the effect of the intervention on drug use was mediated by decision-making skills. Conclusions: The #Tamojunto program increased drug use through decreasing decision-making skills. The findings demonstrate that this program changes decision-making skills but in the opposite direction proposed by the theoretical model of the program, suggesting that modifications are needed to produce the intended effect of the program.