Intervention Characteristics in Relation to School-Based Mental Health Service Outcomes: A Secondary Meta-Analysis

Background and Purpose

A recent meta-analysis of school-based, teacher-delivered mental health interventions (Franklin et al, 2017) found that teacher delivered interventions were overall effective in schools. However, that study presented limited results on interventions’ characteristics in relation to treatment effect sizes. Using a secondary meta-analysis approach, this present paper conducts secondary moderator analysis examining if any intervention characteristic (e.g., modality, tiers of intervention, duration) moderates effect sizes. Results of the study will provide important information that will guide school mental health professionals, school administrators and policy makers in the future design of teacher-delivered mental health interventions in schools.

Methods

The meta-analysis followed the Cochrane Collaboration Guidelines (Higgins & Green, 2011) and searched across nine electronic databases, 19 intervention websites, and articles’ reference lists for studies published from 2000 to the September of 2016. Four primary search terms were used across electronic databases and websites: “school*,” “intervention*,” “random*,” and “teacher*” for full-text search. The dataset included 123 effect size estimates from 24 primary studies and included randomized-control-trial (RCT) studies that examined the clinical effectiveness of a teacher-delivered, school-based mental health interventions for students’ internalizing and/or externalizing outcomes. Single-predictor moderator analysis is performed by entering moderators into the meta-regression model once a time with effect size estimate as the outcome variable.

Results

Using a random effects model, the overall intervention effect was statistically significant for internalizing behaviors, d = 0.13, p = 0.02, but not for externalizing behaviors, d = 0.03, p = 0.23. For internalizing outcomes, tiers of intervention significantly moderated treatment effect size. Classroom-based interventions in primary studies reported significantly greater treatment effect for internalizing behaviors than small-group-based interventions, = -0.30, p = 0.02. Subgroup analysis indicated an overall significant intervention effect for internalizing outcomes across classroom-based interventions, d = 0.22, p < 0.001, but not across small group-based interventions, d = -0.08, p = 0.25. For intervention effects on internalizing behaviors, tiers of intervention significantly moderated intervention effect. Classroom-based interventions in primary studies reported significantly greater intervention effect for internalizing behaviors than small group-based interventions, = -0.30, p = 0.02. Subgroup analysis indicated an overall significant intervention effect for internalizing behaviors across classroom based interventions, d = 0.22, p < 0.001, while the overall effect of small group based interventions for internalizing behaviors was not significant, d = -0.08, p = 0.25. In addition, type of measurement used in primary studies moderated treatment outcomes of internalizing behaviors. Outcomes evaluated using standardized measures were associated with significantly greater intervention effect than those measured by behavioral observations or a combined approach, = -0.50, p < 0.01.

Conclusions and Implications

This secondary meta-analysis reports results on how various intervention characteristics affect the treatment effects on outcomes of teacher-delivered school-based mental health interventions. The primary overall result indicates that the variables of Tiers of the intervention and the evaluation method significantly moderated treatment outcomes for internalizing behaviors. For externalizing disorders, the overall treatment effects were not significant. This calls attention for both researchers and practitioners to further improve services for externalizing outcomes.