A randomized iterative approach to optimizing an online substance use intervention for collegiate athletes

Background: Rates of drug use among collegiate athletes are high, yet there are few evidence-based interventions for this population. myPlaybook, an online intervention for collegiate athletes, targets multiple predictors of drug use (i.e., norms, positive and negative expectancies about use, and harm prevention intentions). Purpose: We aimed to optimize modules from myPlaybook. Method: We evaluated modules through three sequential randomized factorial trials, using the Multiphase Optimization Strategy framework. We recruited and randomized 54 (Trial 1), 47 (Trial 2), and 42 (Trial 3) schools and invited all first-year and transfer collegiate athletes to participate. Athletes completed a baseline survey, their randomly assigned modules, and immediate posttest and 30-day follow-up surveys. Across trials, 3,244 (48.8% female), 2,837 (51.9% female), and 2,193 (51.4% female) athletes participated. In Trial 1, we evaluated and revised less effective modules (defined as d < 0.3-0.4 for targeted outcomes). In Trial 2, we re-evaluated and revised less effective modules. In Trial 3, we re-evaluated the revised modules. Results: Trial 1: All effects were d < 0.15, so we revised modules to target proximal outcomes (i.e., the hypothesized mediating variables in our conceptual model), rather than specific drug use behaviors. Trial 2: Most effects were d < 0.3, so we revised all modules. Trial 3: The norms module improved descriptive and injunctive norms (all d >0.35). The expectancies module improved alcohol positive expectancies (d = 0.3). The other modules were not effective. Conclusions: After three trials, two myPlaybook modules substantially improved proximal outcomes, increasing the likelihood that the combined intervention will have a meaningful clinical impact on collegiate athletes' drug use.