The Effectiveness of Dialectical Behavior Therapy in Reducing Self-Directed Violence and Suicidality in Usual Care Settings: A Meta-Analysis

Abstract

Dialectical behavior therapy (DBT) has shown efficacy in reducing suicidality, though less is known about its effectiveness when delivered in usual care (UC) settings, where the resource-intensive nature of the program often results in modifications during implementation. This systematic meta-analysis examined the effectiveness of DBT in reducing suicidality-related outcomes when delivered in usual care settings. A systematic search was conducted and studies were identified for inclusion if they evaluated a DBT-based intervention with original, quantitative data reflecting pre- and post-treatment measurement of at least one of the three suicidality-related outcomes: (i) self-directed violence (SDV), (ii) crisis service use (CSU), and (iii) suicidal ideation (SI). Random effects models were used to estimate pooled effects. Meta-regression analyses were used to compare pooled effects across usual care and non-usual care settings and to examine the differential effects of theoretical predictors often involved in treatment modification. The final sample included 83 studies. In uncontrolled studies, DBT was associated with a large reduction in SDV and medium reductions in CSU and SI. In controlled studies, DBT demonstrated a significant effect for SDV but not for CSU or SI. Adjustment for publication bias attenuated the CSU findings. No differences were observed between usual care and non-usual care settings. Substantial heterogeneity was observed across models. In uncontrolled studies, positive associations emerged between therapist training and SDV reduction, adherence reporting and CSU reduction, and outpatient setting and SI reduction. Results provide the strongest support for the effectiveness of DBT in reducing SDV, with more limited and variable evidence for CSU and SI. Results should be interpreted with caution given heterogeneity and evidence of publication bias.