Table 1.

Interventions for adolescent self-harm with family component

Intervention, sample	Sessions	Evidence available	Advantages	Disadvantages	Limitations
Brief single session family interventions
Therapeutic assessment (TOTAL)13 12–18 years, n=69	Single session Outcome assessment: 1 year, 2 years	RCT Control: UC	Improved engagement at 1 year and 2 years follow-up (IRR 1.67 (95% CI 1.22 to 2.28), z=3.22, P=0.001)	No significant effect on number of presentation to ED or self-harm episodes	Serious risk of bias: performance and detection bias due to difficulty in blinding clinical personnel due to nature of intervention, reporting and attrition bias unclear
Family-based CBT14 (family intervention for suicide prevention, FISP) 10–18 years, n=181	Single session CBT in ED, phone contact 48 hours post-discharge and at other times during 1st month	RCT Control: EUC	Engagement with the outpatient services better in the FISP group as c/t control group at 2 months f/u (92% vs 76%; OR=6.2; 95% CI=1.8 to 21.3, P=0.004)	No significant effect on SI or SA.	Serious risk of bias: performance and detection bias due to difficulty in blinding clinical personnel due to nature of intervention
Family-based crisis intervention15 (FBCI) 13–18 years, n=100	Single session based on CBT techniques Outcome assessment: 1 day, 1 week, 2 weeks, 1 month, 3 months	Pilot study	Suicidal adolescents and families presenting to ED during FBCI  significantly more likely to be discharged home (and not admitted to an inpatient psychiatry unit) as compared with the comparison cohort (65% vs 44.7%)	No improvement in depression, hopelessness, family cohesion or adaptability at various time points. No data available on SI or SA.	Very serious risk of bias: no randomisation, no blinding of participants and clinical personnel resulting in performance and detection bias, reporting bias was unclear.
Intermediate-level family interventions
Resourceful Adolescent Parent Programme (RAP-P)16 12–17 years, n=48	Four sessions 2 hours each delivered weekly/fortnightly. Sessions included: Outcome assessment: baseline, 3 months, 6 months	RCT Control: UC	Significant reduction in adolescent suicidality (included suicide attempt, ideas, intent, other deliberate self-harm behaviour) with intervention post-treatment and at 6 months f/u (ES 0,72), mediated by improvement in family functioning. Reduced psychiatric disability (the effect size not specified).	No differentiation between those with suicidal ideation and attempters	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention.
Brief home-based problem-solving treatment17 <16 years, n=162	One assessment session in the hospital Four sessions in home by social worker Outcome assessment: baseline, 2 months, 6 months	RCT Control: UC	Significant improvement in SI at 2 (df=2.49, F=8.7, P<0.01) and 6 months (df=2.48, F=8.6, P<0.01) in non-depressed group.	Only measured SI. No outcome measures related to SA or NSSI. No significant improvement in SI overall (mean difference between groups at 2 months: − 3.37 (95% CI − 19.3 to − 12.5); 6 months: −5.1 (CI − 17.5 to −7.3)) and in depressed subgroup. No significant improvement on the measures of hopelessness or family functioning.	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention, reporting bias was unclear.
Safe Alternatives for Teens   and  Youth  Programme (SAFETY)18 11–18 years, n=42	Family centred treatment delivered by two therapists, one for youth and another for the family, over 12 weeks. Outcome assessment: baseline, 3 months (or end of treatment), 6 months, 12 months	RCT Control: E-TAU	Significant between group difference with less SA in intervention group at 3 months axe (z=2.45; P=0.01, NNT=3.0).	Nil significant difference in NSSI between two groups. Treatment effects weakened on 6 and 12 months follow-up (the values at these time points not reported).	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention, reporting bias unclear.
Attachment-based family therapy19 12–17 years, n=66	Intervention for 12 weeks: relational reframe task, adolescent alliance task, parent alliance task, reattachment task, competency task, enhanced usual care. Facilitated referral process with ongoing clinical monitoring. Outcome assessment: baseline, 6 weeks, 12 weeks, 24 weeks	RCT Control: EUC	Significant reduction in self-rated and clinician rated SI at treatment end (12 weeks, effect size 0.95 in favour of ABFT) and follow-up period (24 weeks, effect size=0.97 in favour of ABFT).	Cannot be used if family therapy contraindicated (eg, if the conflict too high). 3/4th sample consisted of African-American, and ½ below poverty level. Findings need replicated in culturally diverse population.	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention, reporting bias unclear.
Intensive family interventions
Multisystem family therapy22 10–17 years, n=156	Intensive treatment for 3–6 months with almost daily contact Outcome assessment: baseline, 4 months, 16 months (1-year post-treatment completion)	RCT Control: inpatient admission	Improvement in SA in those aged 9–12 years post-treatment (4 months) and 1 year f/u as reported by youth and caregivers (intergroup difference- control vs intervention 50% vs 20% reported SA at 4 months, 20% vs 10% at 1 year f/u). In 12–17 years, significant improvement in SA post-treatment & 1 year follow-up only on youth report (TE 3.6, not supported by caregiver reports). Improvement in parental control at 1 year f/u (TE 2.08)	No significant difference in SA in 12–17 years at 1 year f/u as reported by caregivers. No improvement in depression, hopelessness and SI post-treatment or post-follow-up.	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention.
DBT-A20 12–18 years, n=77.	DBT 19 weeks Two sessions/week (individual and multifamily skills training each week), family therapy sessions, telephone coaching Outcome assessment: baseline, 9 weeks, 15 weeks, 19 weeks	RCT Control: EUC for 19 weeks (one session per week)	Significant decrease in self-harm frequency during and post-treatment (at 19 weeks) on longitudinal analyses (P=0.021) and depressive symptoms (ES=0.88). Significant improvement in SI (ES=0.89) and hopelessness (ES=0.97).	No significant difference in cross-sectional analyses of number of self-harm episodes per patient by the end of the therapy period or depression scores. No information available on maintenance of treatment effect. Follow-up was restricted to post-treatment period.	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention, reporting bias unclear.
DBT-A25 13–19 years, n=29	26 weeks, weekly individual therapy and group skills training, telephone support Outcome assessment: Baseline, 3 months, 6 months	RCT Control: TAU	DBT-A was acceptable to clients, parents, caregivers and clinicians; 93% participants in the intervention group completed the therapy.	No significant reduction in the proportion of participants repeating SH or the numbers of SH episodes per patient by the end of the therapy period in intervention group.	Serious risk of bias: performance and detection bias due to difficulty in blinding participants and clinical personnel due to nature of intervention; reporting bias unclear.
DBT adaptation for adolescents24 n=111	12 weeks (twice a week with individual, once-a-week family skills training, family therapy if needed) Outcome assessment: baseline, 12 weeks	Pilot study for intervention adaptation	Significant reduction in number of hospital admissions and treatment completion rates as c/t control (0% in DBT group as c/t 13% in control group). Significant improvement in SI and psychiatric symptoms in the treatment group when compared with pretreatment period (SI = 9.8 pretreatment, 3.8 post-treatment, P<0.05).	No significant improvement in SA with treatment. DBT group consisted of adolescents meeting three or more borderline personality criteria with SA in past 16 weeks and might differ in phenomenology to control group that had either of two criteria.	Very serious risk of bias: selection bias due to non-randomised study design, difficulty in blinding of participants and clinical personnel due to nature of intervention leading to performance and detection bias.
MBT adaptation for adolescents21 14.8 years, n=80	1 year (weekly MBT-A, monthly MBT-F) Outcome assessment: baseline, 3 months, 6 months, 9 months, 12 months Interviews: baseline, 1 year	RCT control: TAU	Significantly more reduction in self-harm on self-report and interview at 1 year in the intervention group as c/t control (56% vs 83%, P=0.01, NNT=3.66, 95% CI=2.19 to 17.32) Significantly greater reduction in self-reported features of borderline personality at 1 year in intervention group as c/t control (d=0.36). Difference in the depressive symptoms between the two group was significant (greater reduction in intervention group, maximum at 9 months, OR  0.21 (95% CI 0.05 to 0.98)	Comparable in quantity Control intervention non- manualised Effect size modest	Serious risk of bias: due to difficulty in blinding of clinical personnel due to nature of intervention leading to performance and detection bias; attrition and reporting bias unclear.

c/t, compared with; CBT, cognitive behaviour therapy; DBT, dialectical behaviour therapy; ED, emergency department; ES, effect size; E-TAU, enhanced treatment as usual; EUC, enhanced usual care; f/u, follow-up; IRR, incidence rate ratio; MBT-A, metallisation-based therapy individual sessions; MBT-F, metallisation-based family therapy sessions; NNT, number needed to treat; NSSI, non-suicidal self-injury; RCT, randomised controlled trial; SA, suicide attempt; SI, suicidal ideation; TE, treatment effect; TAU, treatment as usual; UC, usual care.