A randomized controlled trial was conducted with 6391 7th- and 8th-grade students in 72 public schools in six Brazilian cities. A confirmatory factor analysis was performed to create a unidimensional model for validity. Linear regression was used to evaluate how baseline past-month drug use, violence, and sociodemographic characteristics affected violence at follow-up. The results showed that previous involvement in violent episodes at school (β = 0.397, p < 0.001) as well as the use of inhalants at baseline (β = 0.358, p = 0.026) predicted higher levels of violence 9 months after baseline. Additionally, girls showed lower levels of involvement in violence than boys (β = −0.144, p < 0.001). School prevention programs and policies should simultaneously address drug use and violence. Inhalant use seems to act as a proxy of future involvement in school violence, independent of previous involvement in violent episodes. Trial registration: RebEC, RBR-4mnv5g. Registered 3 July 2014, http://www.ensaiosclinicos.gov.br/rg/RBR-4mnv5g/.

在巴西六个城市的72所公立学校中对6391名7年级和8年级学生进行了随机对照试验。进行了验证性因素分析以创建有效性的一维模型。线性回归用于评估过去一个月的基线吸毒，暴力和社会人口统计学特征如何影响随访中的暴力。结果表明，以前参与学校暴力事件（β  = 0.397，p  <0.001）以及基线使用吸入剂（β  = 0.358，p  = 0.026）预测基线后9个月暴力行为水平会更高。此外，与男孩相比，女孩的暴力行为水平较低（β  = −0.144，p <0.001）。学校预防计划和政策应同时解决吸毒和暴力问题。吸入剂的使用似乎是未来参与学校暴力的代名词，与先前参与暴力事件无关。试用注册：RebEC，RBR-4mnv5g。于2014年7月3日注册，http：//www.ensaiosclinicos.gov.br/rg/RBR-4mnv5g/。