Does treatment method matter? A meta-analysis of the past 20 years of research on therapeutic interventions for self-harm and suicidal ideation in adolescents

Abstract

Background: Self-harm is a clinically relevant and prevalent behaviour which peaks in adolescence. Given the high prevalence of self-harm, the high levels of psychiatric comorbidity, and its role as a risk factor for suicide, delivering evidence-based care is critical.

Methods: We conducted a systematic review and meta-analysis of the literature on treating self-harm in adolescents (12-19 years) published in the last 20 years, identifying 25 randomised controlled trials. We calculated the effect of treatment interventions relative to active control conditions in reducing self-harm, suicidal ideation and depressive symptoms.

Results: Overall, treatment interventions fared slightly better than active controls in decreasing self-harm (d = 0.13, 95% CI 0.04-0.22, p = .004), suicidal ideation (d = 0.31, 95% CI 0.12-0.50, p = .001) and depressive symptoms (d = 0.22, 95% CI 0.07-0.38, p = .006). Subgroup analysis of specific therapies revealed moderate effects of DBT-A in reducing self-harm (d = 0.51, 95% CI 0.18-0.85, p = .002) and suicidal ideation (d = 0.48, 95% CI 0.17-0.80, p = .003), as well as moderate effects of family-centred therapy in the treating suicidal ideation (d = 0.58, 95% CI 0.01-1.15, p = .049).

Conclusions: The findings of our meta-analysis indicate that, overall, currently available treatments are effective in treating self-harm, suicidal ideation, and depressive symptoms in adolescence. Although the treatment intervention conditions showed only small to moderate effects in comparison to active controls, these differences were statistically significant and are clinically important. Further research is needed to understand the reduction in self-harm within active controls, which may arise due to the natural course of self-harm, or the potential efficacy of treatment as usual and enhanced usual care. Given the significant reduction of self-harm in active control conditions, delivering effective care to a large number of adolescents with self-harm may require developing stepped-care models in clinical practice. Expensive and poorly available treatments should be targeted at young people who most need them.

Keywords: Adolescence; Depression; NSSI; Nonsuicidal self-injury; Self-harm; Self-injury; Suicidal behaviour; Suicidal ideation.

Fig. 1

PRISMA flowchart showing the screening, exclusion, and inclusion criteria

Fig. 2

Forest plot of trials comparing the effect of therapeutic interventions and controls on self-harm. Note: Displays the standardized mean difference (Cohen’s d) in post-treatment self-harm, a positive effect size indicates that the outcome was in favour of therapeutic interventions. The average effect was calculated using a random-effects model

Fig. 3

Forest plot of trials comparing the effect of therapeutic interventions and controls on suicidal ideation. Note: Displays the standardized mean difference (Cohen’s d) in post-treatment suicidal ideation, a positive effect size indicates that the outcome was in favour of therapeutic interventions. The average effect was calculated using a random-effects model

Fig. 4

Forest plot of trials comparing the effect of therapeutic interventions and controls on symptoms of depression. Note: Displays the standardized mean difference (Cohen’s d) in post-treatment symptoms of depression, a positive effect size indicates that the outcome was in favour of therapeutic interventions. The average effect was calculated using a random-effects model

Fig. 5

Effect sizes of the difference between therapeutic interventions and control groups by the time of measurement in self-harm. Note. The radius of the points is drawn proportional to the inverse of the standard errors (i.e., studies with greater statistical power are shown as larger points)

Fig. 6

Funnel plots for (a) self-harm, b suicidal ideation and (c) depressive symptoms, showing limited evidence for publication biases across the three outcomes