RESUMO
BACKGROUND: During adolescence, suicide risk increases; effective treatments are needed to reduce risk. METHODS: Databases were searched (1995-2020) for randomized controlled trials (RCTs) concerning psychosocial treatments for suicide prevention in adolescents (10-18 yrs). Data were extracted from the timepoint closest to 6 months. Cohen's ds were estimated for reducing suicidal ideation (SI), self-harming behaviors (SHB) excluding strictly non-suicidal self-injury, and suicide attempts (SA) and analyzed using generalized least square regression. Meta-analytic innovations included within-person correlations to reflect trait suicidality; annualization to control for exposure; estimated lifetime risk based on ages; and modeling inclusion/exclusion criteria. Alternate approaches included relative risk and comparison of intervention and control treatments to baseline. RESULTS: Of 30 RCTs, 6 assessing SHB (4 measuring SA), and 7 assessing SI demonstrated treatment effectiveness. Overall, interventions decreased SI (n = 25) with low effect size (d = 0.08, p = 0.01), non-significant after controlling for publication bias (d = 0.05, p = 0.1); interventions were non-significant for SHB (n = 25, d = 0.001, p = 0.97) or SA (n = 18, d = 0.03, p = 0.52). To prevent one SHB, the number needed to treat (NNT) was 45[26,156]; for SA, NNT=42[24,149]. Non-superiority may relate to effectiveness of control treatments. Thus, experimental and control treatments also were compared to baseline: both reduced SI (p < 0.0001), and effectiveness improved for SHB (NNT=12) and SA (NNT=11). LIMITATIONS: Study heterogeneity and inconsistent statistical reporting limited meta-analysis. CONCLUSIONS: Psychosocial interventions for suicide risk in adolescents showed little effectiveness compared with control treatments; suicide outcomes improved in both groups compared to baseline. Different approaches may be needed, including precision medicine methodologies and standardized statistical reporting criteria.
