Testing Mantente Real, the Culturally Adapted Version of Keepin' It Real for Mexico: Effects on Long-Term Substance Use Trajectories Using Repeated Measures Latent Class Analysis

Background and Purpose: Few evidence-based programs are available to address increasing adolescent substance use rates in Mexico. To fill this gap a bi-national team culturally adapted the keepin’ it REAL (kiR) adolescent substance use prevention intervention, and tested it in an RCT in Mexico’s largest cities. KiR is an efficacious, cost-effective curriculum for middle school students shown to prevent substance use in multiple countries and continents. A multiphase adaptation tailored kiR’s core prevention elements—drug resistance training, risk assessment, and communication skills—to urban Mexico’s cultural and social context. It addressed gendered connections between substance use and violence perpetration and victimization and added new videos shot in recognizable city locations. This presentation reports the adapted intervention’s effects on trajectories of use of different substances from pre- to later adolescence.

Methods: A stratified probability sample of 36 public middle schools in Mexico City, Guadalajara and Monterrey, 12 per city, was randomized to three conditions: Culturally adapted kiR (kiR-A), Original kiR (US version) translated into Spanish (kiR-O), and treatment-as-usual Control. Regular teachers with training in kiR implemented the curricula over 3-4 months. Seventh-grade students with parental consent completed pretests in Fall 2017 (n=5,524), and three posttests: later in 7^th grade (T2), in 8^th (T3) and 9^th grade (T4). Patterns of last 30-day use (versus non-use) of five substances (alcohol, tobacco, inhalants, marijuana, other “hard” drugs) were analyzed from T1 through T4. The effectiveness of kiR-A and kiR-O compared to Controls was analyzed using repeated measures latent class analysis (RM-LCA) in Mplus. Models employed the R3STEP, adjusted for attrition using FIML estimation, and accounted for city- and school-level random effects (clustering).

Results: RM-LCA produced five distinct trajectories of recent substance use: (1) non-users, sustained non-use of all substances from pretest through T4 (49%); (2) initiators of licit use only after pretest (alcohol, tobacco) (28%); (3) initiators of both licit and illicit use (12%); (4) reduction/cessation, those reporting rapidly declining use of all substances (8%); and (5) maintenance of pretest patterns of polysubstance use (3%). In models using intervention conditions as predictors of latent class assignment, kiR-A students were more likely than Controls to be in the declining substance use class, and less likely to be in the two initiating classes where use increased sharply. Unexpectedly, kiR-A students were also slightly more likely than Controls to remain in the polysubstance-using class. The latent class assignments for un-adapted kiR-O students showed no significant differences with Controls.

Conclusions and Implications: Unlike most tests of efficacy for substance use programs that focus on discrete effects on single substances, RM-LCA provides a comprehensive portrayal of substance use trajectories in combination and over time. The findings suggested that the culturally adapted version of kiR for Mexico helped students avoid, delay or reverse until much later a normative pattern in early adolescence of initiating licit and/or illicit substance use. The finding that adapted kiR did not produce desirable changes among pre-existing polysubstance users could indicate that universal prevention programs like kiR need augmentation for this small but highly at-risk number of students.