Using a Multicomponent Implementation Strategy to Increase Adoption and Effectiveness of a Universal Mental Health Prevention Program in Australian Primary Schools: a Cluster Randomized Trial Using a Type-3 Hybrid Design

It has been demonstrated in multiple randomized trials that the PAX Good Behavior Game is an effective school-based program for reducing the severity and progression of childhood emotional and behavioral problems that are prognostic of mental disorders, personality disorders, and self-harm behavior in adolescence and early adulthood. Embedding effective programs into routine practices of frontline settings is a global priority for mental health prevention, however, little is known about what strategies effectively support program implementation in schools. The purpose of this study was to test the effect of a multicomponent implementation strategy on the adoption and efficacy of the PAX Good Behavior Game, using a cluster randomized controlled trial with a type-3 implementation-effectiveness design. The trial was conducted in 25 primary schools across New South Wales, Australia. Schools were randomly assigned to receive the PAX Good Behavior Game plus an implementation toolkit to support adoption (intervention group) or the PAX Good Behavior Game only (control group). The primary outcome was change in rate of program adoption, measured as current use or support of the PAX Good Behavior Game, first measured at 6 weeks post-registration (T0). The secondary effectiveness measure was change in students' emotional and behavioral problems, using the teacher-reported Strengths and Difficulties Questionnaire (SDQ). The first measure of the SDQ occurred at trial registration, prior to program implementation (T0). Both adoption and effectiveness were re-measured at 6 months post-registration (T1). All data were analyzed using intention-to-treat methods. Early program adoption was higher in the intervention group compared to control group (93.6% vs 45.2%: OR = 21.20, 95% CI [3.50, 128.45], z = 3.32, p < 0.001), with no differential effects at 6 months (96.8% vs 95.7%: OR = 1.21, 95% CI [0.16, 9.04], z = 0.18, p = 0.855). Emotional and behavioral problems reduced in both groups from baseline to 6 months (p < 0.001, d = - 0.25), with no differential effects observed at T1 (p = 0.474, d = - 0.08). There was a positive linear relationship between adoption and the perceived acceptability, appropriateness, and feasibility of the program for the pooled sample. It seems that providing implementation support early in the process of establishing new innovations in schools may help accelerate early adoption by increasing perceived appropriateness and acceptability, but uncertainty remains as to what support is needed to optimize implementation and effectiveness longer term. The protocol was retrospectively registered with the Australian New Zealand Clinical Trials Registry, ACTRN12621001125819.