Evidence-Based Collaboration with Teachers: Systematic-Review and Meta-Analysis of Teacher-Delivered-School-Based Mental Health Interventions

Background: Many mental health challenges begin in childhood with approximately 1-in-5 adolescents having a diagnosable mental disorder, yet most do not receive or seek service due to various perceived barriers. This disparity is amplified amongst racial minority and low-income students. School social workers increasingly partner with teachers to deliver school psychosocial interventions, across Tiers 1, 2, and 3, with the aim of increasing mental health services for students of color and low income students. Current research reports mixed findings concerning the effectiveness of interventions delivered by teachers. The purpose of this systematic review and meta-analysis is to examine the effectiveness of teacher-delivered, school-based mental health interventions with students from different sociodemographic backgrounds and who meet criteria for internalizing and/or externalizing behavioral outcomes.

Methods: Following the Cochrane-Collaboration-Guideline, this study reviewed all published and unpublished studies of teacher-delivered school-based mental health interventions. We used four primary search terms: school*, *intervention*, *random*, *teacher*, across nine databases. Twenty-four studies met inclusion criteria and were coded for type of intervention, tier of intervention, sociodemographic and other variables by 4 independent coders. Small sample size corrected Hedges’s g effect size estimates were calculated and Robust-Variance-Estimation (RVE) in meta-regression was used to synthesize effect sizes and conduct moderator analysis. Compared to other statistical procedures that also handle within-study dependence (e.g., multilevel meta-analysis), RVE fits better with our existing data because it makes no assumptions about effect sizes’ sampling distributions and can estimate the covariance structure of the dependent effect sizes without actually knowing it. This study also assessed publication bias and conducted a risk of bias rating.  

Results: Twenty-four studies (N=32,985) were synthesized in the meta-analysis. Eighteen studies included interventions with multiple components using a combination of cognitive-behavioral, social-skills-training, talk therapy and peer mediation. Twenty studies used teachers as the primary interventionist, and 10 studies used a Tier-1 intervention. Treatment effect size for internalizing outcomes was d=.133, 95% CI [.002,.263], and d=.015, 95% CI [-.037,.066] for externalizing outcomes. Difference between the mean treatment effect size between internalizing and externalizing outcomes was statistically significant, d=.118, 95% CI [.034,.202]. Proportion of males was negatively associated with treatment effect, b=-.017, p<.05 for internalizing disorders, and higher percentage of Caucasian participants is significantly associated with positive treatment outcome, b=.002, p<.05, for externalizing disorders. Tier-1 interventions were more effective than other tiers of intervention (b=.126, p=.06). No presence of publication bias was found using funnel plots.

Conclusions: This study builds on school mental health literature and found Tier-1 interventions delivered by teachers are more effective with internalizing than externalizing outcomes. Overall treatment effects were moderated by both race and gender; these interventions are more effective with female for internalizing outcomes and with Caucasians for externalizing outcomes. This review points to the need to give more attention to intervention implementation and report on the quality of teacher supervision to help gain a better understanding for the implementation factors that may influence outcomes, especially for Tier 2 and 3 interventions that may require greater effort by teachers and the learning of new skills.