The treatment of suicidality in adolescents by psychosocial interventions for depression: A systematic literature review

Transcription

1 627374ANP / ANZJP ArticlesDevenish et al. research-article2016 Key Review The treatment of suicidality in adolescents by psychosocial interventions for depression: A systematic literature review Australian & New Zealand Journal of Psychiatry 2016, Vol. 50(8) DOI: / The Royal Australian and New Zealand College of Psychiatrists 2016 Reprints and permissions: sagepub.co.uk/journalspermissions.nav anp.sagepub.com Bethany Devenish 1, Lesley Berk 1,2 and Andrew J Lewis 1,3 Abstract Objective: Given depression is a risk factor for suicidal behaviour, it is possible that interventions for depression may also reduce the risk of suicide in adolescents. The purpose of this literature review is to determine whether psychological interventions aimed to prevent and/or treat depression in adolescents can also reduce suicidality. Methods: We conducted a systematic review of psychological interventions aimed to prevent and/or treat depression in adolescents in which outcomes for suicidality were reported, using five databases: PsycINFO, Embase, Medline, CINAHL and Scopus. Study quality was assessed using the Cochrane Collaboration s tool for assessing risk of bias. Results: A total of 35 articles pertaining to 12 treatment trials, two selective prevention trials and two universal prevention trials met inclusion criteria. No studies were identified that used a no-treatment control. In both intervention and active control, suicidality decreased over time; however, most structured psychological depression treatment interventions did not outperform pharmaceutical or treatment as usual control. Depression prevention studies demonstrated small but statistically reductions in suicidality. Limitations: Analysis of study quality suggested that at least 10 of the 16 studies have a high risk of bias. Conclusive comparisons across studies are problematic due to in measures, interventions, population and control used. Conclusions: It is unclear whether psychological treatments are more effective than no treatment since no study has used a no-treatment control group. There is evidence to suggest that Cognitive Behavioural Therapy interventions produce reductions in suicidality with moderate effect sizes and are at least as efficacious as pharmacotherapy in reducing suicidality; however, it is unclear whether these effects are sustained. There are several trials showing promising evidence for family-based and interpersonal therapies, with large effect sizes, and further evaluation with improved methodology is required. Depression prevention interventions show promising short-term effects. Keywords Suicide, adolescents, depression, systematic review, treatment Background The World Health Organization s (WHO, 2014) first world suicide report identifies suicide as the second leading cause of death in 15- to 29-year-olds. Suicide behaviour consists of death by suicide (self-inflicted and intentional killing of oneself), suicide attempts (self-injurious behaviour with the intent to cause death) and suicidal ideation (persistent, serious thoughts of suicide). Such behaviour can be conceptualized as a continuum, and here, we refer to all these 1 Faculty of Health, School of Psychology, Deakin University, Burwood, Australia 2 Department of Psychiatry, Faculty of Medicine, Dentistry and Health Sciences, Deakin University, Burwood, Australia 3 School of Psychology and Exercise Science, Murdoch University, Murdoch WA, Australia Corresponding author: Andrew J Lewis, School of Psychology and Exercise Science, Murdoch University, 90 South Street, Murdoch WA, 6011, Australia. a.lewis@murdoch.edu.au

2 Devenish et al. 727 features with the term suicidality (Marušič, 2004). While suicidality is rare in childhood, it increases in frequency across adolescence and shows an increasing rate from 15 until around 24 years of age (Australian Bureau of Statistics, 2013). Risk factors for suicidality in adolescents are complex and include previous suicide attempts, psychopathology, peer victimization, a history of sexual or physical abuse, self-harm, social isolation, poor problem-solving and coping skills, low self-esteem, family dysfunction, repeated exposure to violence and ease of access to a means of suicide (e.g. a lethal weapon) (Miller and Eckert, 2009). Additionally, there are strong theoretical and empirical links between suicidality and depression (Miller and Eckert, 2009; Tuisku et al., 2014). Direct intervention for adolescent suicidality Psychological interventions designed specifically for the treatment of suicidality in adolescents have not demonstrated superior efficacy compared to usual care and medication (Corcoran et al., 2011). This was highlighted in a 2011 meta-analysis of 17 interventions designed to treat suicidality in adolescents (Corcoran et al., 2011). This review found that adolescents who received a suicide focused intervention were slightly less likely than controls to experience suicidal ideation and suicidal events at posttest (Corcoran et al., 2011). However, when followed up 4 6 months after treatment, adolescents who had received the active intervention were more likely to experience suicidal events, but less likely to report suicidal ideation (Corcoran et al., 2011). The authors concluded that interventions focused specifically on suicidality were at best marginally effective in the short term, but importantly may also lead to an increase in suicidality over the longer term (Corcoran et al., 2011; Cox et al., 2012). An earlier review and meta-analysis of 15 trials evaluated all randomized controlled trials (RCTs) of clinical interventions for young people who presented to a clinical setting with a recent history of suicidal ideation, suicide attempt or deliberate self-harm, and found that individual cognitive behavioural therapy (CBT) interventions demonstrate some promise for the reduction of suicide risk in young people (Robinson et al., 2011). A later systematic review was conducted by the same researchers, of schoolbased postvention, prevention and early interventions for suicide, suicide ideation, suicide attempts or self-harm (intention not specified) (Robinson et al., 2013). This review found that gatekeeper interventions, i.e., interventions designed to improve the capacity of professionals and paraprofessionals to work with suicidal teenagers, showed promise for increasing knowledge and improving attitudes, while screening programmes successfully identified students at risk (Robinson et al., 2013). It should be noted that there is a lack of consensus as to whether non-suicidal selfharm behaviour should be included in reviews that focus on suicidal behaviour (Halicka and Kiejna, 2015). The Robinson review also suggested that there was limited evidence that school-based interventions reduce suicidal behaviour and ideation itself (Robinson et al., 2013). Universal prevention approaches demonstrated some effects, while indicated prevention interventions showed promise (Robinson et al., 2013) and require further evaluation. Suicidality and adolescent depression Psychiatric disorders are present in up to 90% of people who die by suicide (Cavanagh et al., 2003) and depression has been identified as one of the most important risk factors for suicidality in adolescents (Beautrais, 2000; Tuisku et al., 2014). While many adolescents who experience major depression are not suicidal (Reynolds and Mazza, 1994), adolescents who died by suicide were 27 times more likely to have been suffering from major depression (Brent et al., 1993). It has also been found that parents have great difficulty accurately identifying the suicidality of their adolescents when they are diagnosed with depression, which further increases the risk of suicide and lack of access to effective interventions (Lewis et al., 2014a). Given the strong links between depression and suicidality, the effective treatment and prevention of depression may be an important pathway to reducing adolescent suicidality. Among the available treatment options, antidepressant medication for adolescent depression remains controversial since a number of studies have shown antidepressants may induce a temporary increase in suicidal ideation and suicide attempts (Lock et al., 2005; Miller et al., 2014). Psychological treatments may offer a promising alternative for reducing both depression and suicidality in adolescents. Research with adults has shown that improved identification and treatment of depression leads to reduced suicide rates (Mann et al., 2005), and improvements in regional care and optimizing treatment for depressed patients led to a 24% reduction in suicidal acts (Hegerl et al., 2006). It is therefore possible that similar results may be found in adolescent populations. To date, no review has explicitly focused on the extent to which depression focused psychological treatments also reduce adolescent suicidality. Depression prevention and treatment programmes often do not specifically target suicidal behaviour, nor report changes in suicidality as an important secondary outcome. Furthermore, the design of trials in adolescent depression often excludes adolescents at risk of suicide, probably due to the ethical concerns in running such trials (Lewis et al., 2014a). Several previous reviews have touched on the topic of reducing both depression and suicidality in adolescents. For example, a Cochrane review published in 2012 (Cox et al., 2012) of psychological and pharmacological therapies for depression provided a subgroup analysis of three studies examining interventions for adolescent suicide

3 728 ANZJP Articles ideation and reported some evidence that psychological interventions have at least equal effectiveness to pharmacotherapy for reducing suicidal behaviours in the long term. Furthermore, a meta-analysis of psychotherapies for child and adolescent depression included a sub-analysis of six depression interventions studies that included a measure of suicidality (Weisz et al., 2006), and reported that the effect of psychological interventions on suicidality was relatively small (d = 0.18, p = 0.07). The Weisz review noted considerable heterogeneity in the type of control group employed by the different studies, including medication placebos, supportive therapy and a waiting list control. Finally, a more recent literature review of seven studies focused specifically on CBT for the treatment of adolescent depression and its effect on suicidality (Spirito et al., 2011) and concluded that most CBT intervention studies provide comparable reductions in suicidality to family therapy, supportive therapy and pharmacotherapy, but show the most promise for also reducing major depressive disorder (MDD) diagnoses and depressive symptoms. This however conflicted with the Weisz et al. (2006) meta-analysis which did not find any evidence that cognitive treatments are more effective than noncognitive approaches. The current review To date, no review has specifically examined as its primary focus, the efficacy of any type of psychological intervention (treatment or prevention) for adolescent depression and its impact on suicide reduction. The prevention of mental disorders is gaining momentum in the light of recent reviews and meta-analyses showing that preventive interventions, particularly those targeting high-risk adolescents, can be effective in preventing the onset of depressive disorders or reducing an increase in depressive symptoms over adolescence (Jacka et al., 2013; Lewis et al., 2014b; Merry et al., 2011). Thus, unlike previous reviews, we explicitly include the effects on suicidality of interventions designed to prevent adolescent depression as well as psychological depression treatment interventions. This review adopts a systematic approach and also compares the efficacy of a variety of both traditional and emerging psychological treatment approaches. As such, our review seeks to answer four main questions: 1. Do psychological depression prevention and/or treatment interventions reduce suicidality in adolescents to a greater extent than receiving no treatment? 2. Do psychological depression prevention and/or treatment interventions outperform placebo control, such as treatment as usual (TAU)? 3. Do psychological depression treatment interventions outperform other interventions, such as treatment with antidepressant medication? 4. Given the differing approaches to treating depression, are there any specific psychological depression prevention or treatment interventions, or approaches to treatment, that lead to more reductions in suicidality than other psychological depression treatments or approaches? Methods Search strategy An electronic literature search was conducted to search for psychological interventions designed to prevent or treat adolescent depression that also included reports of outcomes on validated measures of suicidality. Searches of title, classification codes, abstract, keyword, subjects and author were conducted examining all studies available up to April 2015 using five electronic databases PsycINFO, Embase, MEDLINE, CINAHL and Scopus. The following search terms were used for all five databases: ([suici*] AND [ behaviour therapy OR behavior therapy OR behavioral therapy OR behavioural therapy OR therap* OR biofeedback OR cognitive therapy OR counsel* OR family therapy OR parent education OR parental education OR parents education OR psychoanalytic therapy OR psychoanalytical therapy OR psychotherapy OR relaxation therapy OR psychosocial OR cognitivebehavior-therapy OR cognitive-behaviour-therapy OR psychoanaly* OR interpersonal therapy OR mindfulness OR mindfulness-based cognitive therapy OR cognitive behavior therapy OR cognitive behaviour therapy ] AND [depressi* OR mood disorder OR mood disorders OR dysthym*] AND [adolesce* OR teen* OR youth*]). Results were limited to peer-reviewed journals. The reference lists of included studies and relevant literature reviews were hand searched to include all relevant empirical studies. In addition, the Australian New Zealand Clinical Trials registry and ClinicalTrials.gov registry were searched for relevant trials, and where further information was required, the authors were contacted. Inclusion and exclusion criteria Studies were included according to three main criteria. First, the sample consisted of adolescents (11- to 19-yearolds) only. While adolescence is often viewed as pertaining to 12- to 18-year-olds, we widened the range for this review as high school ages typically range from 11 to 19 years of age and we did not want to exclude interventions conducted in a high school setting. Any studies which included participants who were younger than 11 or older than 19 years of age were excluded from this review. Studies which were initially conducted with adolescents but also reported on outcomes at a later time point were not excluded since these outcomes were of interest. The primary focus of the review

4 Devenish et al. 729 was on papers which reported on outcomes at the end of the intervention or treatment phase, but all time points were included in order to evaluate whether any changes are retained at follow-up. Second, studies were only included if they included a treatment or prevention intervention that was psychological in nature and contained content designed to reduce symptoms of depression. Content was considered to be psychological in nature if it included content based on a recognized psychological theory, including but not limited to CBT, interpersonal therapy, attachment-based, psychodynamic, family therapy or mindfulness. We did not consider deep brain stimulation techniques such as transcranial magnetic stimulation as being psychological in nature. Content was considered to be designed to reduce symptoms of depression if it was explicitly indicated to be so, or if it included discussion of depression, depression symptoms or management of depression. Finally, studies were only included if they reported a valid measure of suicidal ideation, gestures or attempts as an outcome variable. Non-fatal self-injury was not considered to meet these criteria if no indication of suicidal intent was assessed. Studies were excluded if the article was not written in English or if the authors stated that the sample consisted primarily of adolescents diagnosed with bipolar or borderline personality disorder, as treatment for mood problems in these is disorder specific (Biskin, 2013; Maalouf et al., 2012; Scott et al., 2006). Nevertheless, depression in adolescents has high comorbidity (Small et al., 2008), and so studies which included participants with comorbid (e.g. psychosis) were not excluded in order to increase the clinical utility of the review. Most RCTs have evaluated a version of CBT for depression (Weisz et al., 2006); however, a growing number of other psychotherapy approaches have emerged that target adolescent depression. Pilot studies were included to gauge the possibility of preliminary support for new treatments, especially given the gaps in the literature on later phase clinical trials of different psychotherapeutic approaches. Single-case designs and evaluations using post-treatment data only were excluded. Study selection and evaluation The study selection process is illustrated in the Preferred Reporting Items for Systematic Reviews and Meta- Analyses (PRISMA) flow chart (Moher et al., 2009) presented in Figure 1. In keeping with PRISMA guidelines, a critical analysis of the included studies was conducted (Moher et al., 2009) and is reported in Table 2. Studies were evaluated according to the Cochrane Collaboration s tool for assessing risk of bias (Green and Higgins, 2011), which evaluates the risk of selection bias, performance bias, detection bias, attrition bias, reporting bias and other biases. However, selective reporting was not relevant in this case since studies had to report on a measure of suicidality to have been included. One author (B.D.) extracted the data and conducted the critical analysis, and the other authors (L.B. and A.L.) reviewed the analyses and discrepancies in rating were resolved by discussion. Effect sizes (Cohen s d) were calculated for all studies in which there were statistically results and where published data permitted such calculations. Where relevant, effect sizes comparing treatment to control group were calculated, or, for studies which did not include a control group, effect sizes comparing baseline to postintervention results were calculated. Results General characteristics PsycINFO yielded 1549 articles, Embase 3410 articles, MEDLINE 2283 results, CINAHL 1494 results and Scopus 4088 articles. A total of 35 articles reported on 16 studies that met inclusion criteria, and a summary of study characteristics can be found in Table 1. No studies comparing a depression intervention or prevention condition to a waiting list or no-treatment control group were identified. This is likely to reflect ethical concerns about leaving adolescents experiencing depression or suicidal ideation untreated. Twelve studies were treatment studies and four were designed to be preventative. Nine studies evaluated the efficacy of CBT, and seven studies evaluated the efficacy of other psychological treatments, which included systemic family therapy, attachment-based family therapy (ABFT) and interpersonal therapy. There were few psychotherapy trials with only three studies (Brent et al., 1997; Eggert et al., 2002; Spirito et al., 2015) comparing a psychological intervention to another or even to a different form of the same type of intervention (e.g. CBT for adolescents only, or adolescents and parents together) (Spirito et al., 2015). Notably, only one prevention study assessed whether the effects of a longer intervention differed to that of a shorter intervention (Eggert et al., 1995). Individual treatment was used in six studies (Brent et al., 1997; Goodyer et al., 2007; March et al., 2007; Spirito et al., 2015; Tang et al., 2009; Vitiello et al., 2011), four studies included family-based interventions (Brent et al., 1997; Diamond et al., 2010, 2013; Spirito et al., 2015), two studies included group treatments (Straub et al., 2013, 2014) and two used a combination of both individual treatment and family sessions (Brent et al., 2009b; Melvin et al., 2006). The four prevention studies were conducted in group settings (Eggert et al., 1995; 2002; King et al., 2011; Perry et al., 2014). Inclusion of participants based on depressive criteria and severity of depression varied widely across studies, with six studies only including adolescents who met the criteria for MDD or dysthymia in a structured interview (Brent

5 730 ANZJP Articles Figure1. PRISMA 2009 flow diagram. et al., 2009b; Goodyer et al., 2007; March et al., 2004; Melvin et al., 2006; Spirito et al., 2015; Tang et al., 2009; Vitiello et al., 2011), two studies included only adolescents whose scores on Beck s Depression Inventory indicated moderate or severe depression (Brent et al., 1997; Diamond et al., 2010), two studies used a combination of elevated depression scores and clinical diagnosis of MDD (Straub et al., 2013, 2014), and one study included adolescents who did not necessarily meet criteria but displayed elevated depressive symptoms (Diamond et al., 2013). One study also required a parent to meet criteria for current or past MDD (Spirito et al., 2015). Additionally, three studies excluded adolescents if they were at high risk of suicide or self-harm (March et al., 2004; Melvin et al., 2006; Tang et al., 2009). Risk of bias Results of the analysis of study quality are presented in Table 2. As can be seen in Table 2, the larger and higher quality studies were RCTs examining CBT in comparison to medication. Six studies had a low risk of bias for more than half of the criteria. All remaining studies had an unclear or high risk of bias. Studies of medication versus psychotherapy The highest quality studies were the of Resistant Depression in Adolescents (TORDIA) study (Brent et al., 2008), The for Adolescents with Depression Study (TADS) (March et al., 2006) and the Adolescent

6 Devenish et al. 731 Table 1. Summary of included studies investigating the effect of adolescent psychosocial depression interventions on suicidality. Study or author name Design of study Total sample size Types of interventions compared condition 1 condition 2 condition 3 (and 4) Suicide measure Outcome at post-treatment Trials of antidepressant medication versus psychological therapies TORDIA RCT trial with four TADS RCT, trial with three ADAPT RCT, trial with three Melvin et al. (2006) TASA (Brent et al., 2009b; Vitiello et al., 2009a) RCT, trial with three Controlled Trial, trial with three 334 Medication; vs CBT 327 Medication; vs CBT 208 Medication; CBT; as usual (TAU) New SSRI (n = 85) New SSRI plus CBT (n = 83) Fluoxetine (n = 109) SSRI and routine clinical care (n = 103) Third condition: Venlafaxine (n = 83); Fourth condition: Venlafaxine plus CBT (n = 83) CBT (n = 111) Combination (n = 107) SSRI, routine clinical care and CBT (n = 105) 73 Medication; CBT Sertraline (n = 26) CBT (n = 22) Combination (n = 25) 124 Medication; specialized psychological therapy for suicidal adolescents Medication algorithm (n = 14) Psychotherapy (n = 17) na K-SADS- PL items on suicidal thoughts and Suicidal ideation Combination (n = 93) SIQ-JR 12 weeks changes; no between SIQ-JR 12 weeks changes; fluoxetine only group ly more suicidal ideation than CBT and combination 12 weeks changes; no between K-SADS-PL weeks changes; no between SSI weeks changes; no between Outcomes at follow-up period Effect size for (unless otherwise indicated) (Cohen s d) 24 weeks no between. 72 weeks no between Medication only = 0.54; Meds plus CBT = week maintenance phase fluoxetine only group ly more suicidal ideation than CBT; 1 year naturalistic followup no between 28 weeks no between Med only = 0.54; CBT only = 0.77; comb = 1.00 Suicidal thoughts: med group = 0.88; CBT group = 0.99; Suicidal ideation med group = 0.98; CBT group = months followup no between CBT = 0.34; Med = 0.19; comb = 0.33 (Continued)

7 732 ANZJP Articles Table 1. (Continued) Study or author name Design of study Total sample size Types of interventions compared condition 1 condition 2 condition 3 (and 4) Suicide measure Outcome at post-treatment Family-based and interpersonal therapy trials Brent et al. RCT, 107 CBT; family (1997, therapy; 2004, trial with non-directive 2005) three supportive therapy. ABFT (Diamond et al., 2010; Shpigel et al., 2012) Diamond et al. (2013) Tang et al. (2009) RCT, trial with two Open trial with one condition RCT, trial with two 66 Family therapy; TAU CBT (n = 37) Systemic behaviour family therapy (SBFT) (n = 35) Attachment- Based Family Therapy (ABFT) (n = 35) Enhanced usual care (EUC) (n = 31) Non-directive supportive therapy (NST) (n = 35) SIQ-JR and clinician rated suicidality 12 weeks changes; no between na BSS Post ABFT ly less suicidal ideation 10 ABFT ABFT na na SIQ-JR 12 weeks reduction in suicidal 73 Interpersonal therapy; TAU Interpersonal therapy (n = 35) TAU psychoeducation and supportive counselling (n = 38) na Brief Suicide Risk Behaviour Scale ideation 6 weeks interpersonal therapy ly less suicidal ideation School-based prevention programmes Eggert RCT 105 School-based et al. prevention (1995), prevention programme; no Thompson trial with treatment et al. three (2000) Eggert RCT 341 School-based et al. prevention (2002) prevention programme; usual trial with care three Assessment only (n = 35) Brief schoolbased individual intervention (n = 117) assessment plus 1 semester ITP (n = 36) Coping and Support Training (n = 103) 2 semester ITP (n = 34) Usual care (n = 121) HSQ 9 months pre post changes; no between BSS 10 weeks changes; no between Outcomes at follow-up period Effect size for (unless otherwise indicated) (Cohen s d) 24 weeks no between CBT 1.02; SBFT 1.18; NST 0.73 Between difference = 0.97; difference: ABFT = 0.59, EUC = 0.24 Pre post 2.10 Between difference = 0.84; Pre post: ITP = 0.98, TAU = 0.08 Pre post in each group was minimum of 0.60 (Continued)

8 Devenish et al. 733 Table 1. (Continued) Study or author name Design of study Total sample size Types of interventions compared condition 1 condition 2 condition 3 (and 4) Suicide measure Outcome at post-treatment Outcomes at follow-up period Effect size for (unless otherwise indicated) (Cohen s d) Miscellaneous pilot studies, community-based cluster trials and non-rct trials Spirito RCT 24 CBT Parent et al. adolescent CBT (2015) trial with two (n = 16) King et al. (2011) Straub et al. (2013) Straub et al. (2014) Perry et al. (2014) Quasiexperimental prevention study with one condition Pilot study with one condition Pilot study with one condition Cluster RCT prevention trial with two 1030 Suicide prevention and depression awareness programme 9 Brief CBT group therapy 15 Brief CBT group therapy 380 Psychoeducation; no treatment Suicide prevention and depression awareness programme Brief CBT group therapy Brief CBT group therapy Psychoeducation (n = 207) Adolescent only CBT (n = 8) na Suicide Severity Rating Scale na na Student Involvement in Suicidal Ideation na na item 13 (CDRS-R) 12 weeks changes; no between Post no changes in suicidal ideation Post reduction in suicidal ideation na na (CDRS-R) Post health classes (n = 173) na Six items of MFQ reduction in suicidal ideation Post no changes 24 week maintenance phase no between 3 months reduction in suicidal ideation Pre post parent adolescent CBT = 0.91 Pre post = 0.16 Pre post 0.76 Pre Post months no changes Pre post psychoeducation = 0.004; No treatment = 0.04 CBT: cognitive behavioural therapy; RCT: randomized controlled trial; SSRI: selective serotonin reuptake inhibitor; TADS: for Adolescents with Depression Study; ADAPT: Adolescent Depression Antidepressants and Psychotherapy Trial; TASA: of Adolescent Suicide Attempters Study; na: not applicable, SIQ-JR: Suicidal Ideation Questionnaire Junior, German version of the Children Depression Rating Scale Revised Version (CDRS-R); HSQ: High School Questionnaire: A Profile of Experiences; MFQ: Moods and Feelings Questionnaire; K-SADS-PL: Kiddie-Sads-Present and Lifetime Version, suicide items only; SSI: Scale for Suicidal Ideation; BSS: Beck Scale for Suicidal Ideation; ITP: Interpersonal Psychotherapy. TORDIA refers to studies published by Brent (2009), Brent et al. (2008, 2009b), Emslie et al. (2010), Vitiello et al. (2011). TADS refers to studies published by Emslie et al. (2006), Kennard et al. (2009), March et al. (2009, 2004, 2006, 2007), Simons et al. (2012), Vitiello et al. (2009b). ADAPT refers to studies published by Goodyer et al. (2007, 2008), Wilkinson et al. (2009, 2011).

9 734 ANZJP Articles Table 2. Critical analysis of included studies. Study/cohort Randomization Allocation concealment Blinding of participants and personnel Blinding of outcome assessors Incomplete outcome data TORDIA (Brent, 2009; Brent et al., 2008, 2009b; Emslie et al., 2010; Vitiello et al., 2011) TADS (Emslie et al., 2006; Kennard et al., 2009; March et al., 2004, 2006, 2007, 2009; Simons et al., 2012; Vitiello et al., 2009b) ADAPT (Goodyer et al., 2007, 2008; Wilkinson et al., 2009, 2011) (Brent et al., 1997; Barbe et al., 2004; Bridge et al., 2005) Low Low High Low Low Low Low High Low Low Low Low High Low Low Low Low Unclear Low Low Melvin et al. (2006) Low Low High Low Low ABFT (Diamond et al., 2010; Shpigel et al., 2012) Low Low Unclear High Low Tang et al. (2009) Unclear Unclear Unclear Low Unclear ITP (Eggert et al., 1995; Thompson et al., 2000) Unclear Unclear High Low Low Eggert et al. (2002) Low Unclear High Low High Spirito et al. (2015) Unclear Unclear Unclear Unclear High TASA (Brent et al., 2009b; Vitiello et al., 2009a) High High High Low Low King et al. (2011) High High High Low High Diamond et al. (2013) High High High Low Low Straub et al. (2013) High High High Low Low Straub et al. (2014) High High High Unclear Low Perry et al. (2014) Low High High High High TADS: for Adolescents with Depression Study; ADAPT: Adolescent Depression Antidepressants and Psychotherapy Trial; TASA: of Adolescent Suicide Attempters Study; ABFT: Attachment-Based Family Therapy. Depression Antidepressants and Psychotherapy Trial (ADAPT) (Goodyer et al., 2008), which evaluated whether treating depression with medication, CBT or a combination of both treatments would lead to greater improvements in depression and suicidality. In these and other pharmacotherapy trials, participants were randomized to receive medication, CBT or a combination of both treatments (Brent et al., 2008, 2009b; March et al., 2009; Melvin et al., 2006), or in one study, CBT or a combination (Goodyer et al., 2007). The results across these five studies were mostly consistent in suggesting that suicidality decreased ly in response to treatment across all, with follow-up assessments suggesting these effects were retained over time, in some over a year later (March et al., 2009; Vitiello et al., 2011). However, while suicidality decreased over time, there were not between participants receiving monotherapies of CBT or medication or those receiving combined treatments. The one exception to this was that participants receiving fluoxetine only were ly more likely to report suicidal ideation or suicide risk than participants receiving CBT at 6 weeks (Emslie et al., 2006), 12 weeks (Emslie et al., 2006; March et al., 2004) and 36 weeks (March et al., 2007). By 1-year naturalistic follow-up, there were no between the three on suicidality (March et al., 2009).

10 Devenish et al. 735 To examine whether homework completion in CBT was related to treatment response, secondary analyses were conducted on TADS data, finding that adolescents in the CBT group who had completed a higher percentage of homework also showed ly greater decreases in suicidal ideation and hopelessness. In the combination group, the only relationship was an association between partial homework completion and lowered self-reported hopelessness (Simons et al., 2012). For three trials, the content of the CBT was focused on reducing depressive symptoms, and there did not appear to be specific content focused on the reduction of suicidality (Kennard et al., 2009; March et al., 2006; Melvin et al., 2006), although in the TADS, a manual-based intervention to address suicidality was utilized (May et al., 2007). The TASA study was aimed at treating depression in adolescents who had recently attempted suicide and treatment involved CBT therapy for adult suicide attempters modified extensively to better meet the needs of depressed and suicidal adolescents (Brent et al., 2009b). The modifications combined information was obtained from TADS, TORDIA and Dialectic Behaviour Therapy manuals (Brent et al., 2009b). CBT versus other psychological therapies The efficacy of CBT for the treatment of suicidality in depressed adolescents has also been evaluated by studies comparing CBT to other therapy approaches (Brent et al., 1997; Spirito et al., 2015). Studies have also examined whether individualized treatment or group treatment within a CBT framework can lead to better outcomes for suicidality (Straub et al., 2013, 2014). In the only study to compare CBT to an alternative psychological intervention, adolescents diagnosed with MDD were randomized to one of three : weeks of CBT, systematic behaviour family therapy or non-directive supportive therapy, followed by 2 4 booster sessions (Brent et al., 2008). Systematic behaviour family therapy combines elements of functional family therapy and Robin and Foster s problemsolving model, and therefore involves the identification of dysfunctional behaviour patterns and a focus on family communication and problem-solving (Brent et al., 1997). Non-directive supportive therapy aims to provide therapeutic support while refraining from teaching specific skills, giving advice or setting limits (Brent et al., 1997). There was a reduction in suicidal ideation across all three at all time points (Cohen s d = ), but no between the three treatment approaches, suggesting all treatments had equal or similar efficacy. Uncontrolled pilot studies assessing a brief manualized group CBT intervention for adolescents diagnosed with depression have been conducted in an inpatient and outpatient setting (Straub et al., 2013, 2014). Suicidal ideation was assessed and ly decreased in both studies (Cohen s d = 0.76 and 0.94); however, as there was no control group, efficacy in comparison to other treatment options has not been evaluated. A randomized controlled study of this intervention is currently underway with results expected later in 2015 (J. Straub, personal communication, 5 May 2014). Finally, an RCT of CBT with and without family member participation found that while suicidality decreased ly over time, there were no between adolescents who participated in adolescent-only CBT, and adolescents who participated in a treatment arm which included one-on-one sessions with a therapist, oneon-one sessions for parents with a therapist and conjoint family sessions (Spirito et al., 2015). Efficacy of other psychological treatments There was a small amount of research on the efficacy of other psychological treatments for depression which also reported on suicidality in adolescents. One study assessed whether interpersonal therapy would decrease suicidal ideation in depressed adolescents at risk of suicide (Tang et al., 2009). Adolescents were randomized to 12 sessions of interpersonal therapy over 6 weeks or treatment as usual (TAU) (psychoeducation and supportive counselling). Participants receiving interpersonal therapy showed a ly greater reduction in suicidality post-intervention than participants receiving TAU, with a between- effect size of d = 0.84 in favour of interpersonal therapy, and a effect size of d = 0.98 (Tang et al., 2009). Research has also been conducted to evaluate whether ABFT reduces suicidal ideation and depression in adolescents. ABFT is a manualized family therapy designed to improve family processes associated with depression, including problem-solving skills, affect regulation and organization (Diamond et al., 2010). Adolescents were randomized to receive either 3 months of ABFT or Enhanced Usual Care (EUC) (facilitated referral with ongoing clinical monitoring) (Diamond et al., 2010). ABFT showed a ly greater rate of change than EUC in self-reported suicidal ideation when assessed post-treatment (12 weeks), maintaining this difference at follow-up (d = 0.97), with ly more participants meeting criteria for clinical recovery assessed as a Suicidal Ideation Questionnaire Junior (SIQ-JR) score of less than 13 and a Scale for Suicidal Ideation (SSI) of zero (Diamond et al., 2010). Clinician-rated suicidality was similar between. In all, 47% of participants met criteria for a depressive disorder, and demonstrated a similar reduction in suicidality as participants with only elevated depressive symptoms (Diamond et al., 2010). However, there were issues with retention of participants in the EUC group, and so it is unclear whether effects were due to the treatment approach or simply the increased support and attention offered in ABFT (Diamond et al., 2010). Shpigel et al.

11 736 ANZJP Articles (2012) further analysed these data to examine whether changes in maternal behaviours were related to changes in adolescent suicidal symptoms, and found no relationship. Preliminary research assessing the feasibility of ABFT adapted for use with lesbian, gay and bisexual suicidal adolescents, most of whom were depressed, showed a decrease in suicidality (Diamond et al., 2013). However, there was no control group and the sample size was small, so further research is needed to assess the efficacy of ABFT for suicidality in this population (Diamond et al., 2013). Prevention programmes Prevention programmes include interventions that aim to reduce a known risk factor for a given outcome. In this case, we consider programmes designed to reduce both depressive symptoms and suicidal ideation since both are clear risk factors for suicide. Four studies examined the efficacy of school-based suicide prevention programmes for adolescents which aimed to reduce suicidal ideation as a risk factor for suicide, with three of these finding suicidality reduced between baseline and post-intervention (Eggert, 1995; Eggert et al., 2002; King et al., 2011). Two studies were universal prevention studies (King et al., 2011; Perry et al., 2014), and two studies were selective prevention trials (Eggert et al., 1995, 2002). In the Eggert et al. (1995) study, students were randomized to receive one of three arms of intervention: an assessment, assessment plus a one semester educational programme or assessment plus a two semester programme. The programme was run in small and covered mood management, school performance and attendance, drug education and skills training in self-esteem, decision making, anger, depression, stressmanagement and interpersonal communication. The assessment measured suicide potential and activated social supports through informing parents of the adolescent s needs and connecting adolescents with a school case-manager (Eggert et al., 1995). All three demonstrated a decline in suicidality across time (Eggert et al., 1995). No between the treatment were observed, suggesting that the assessment was just as efficacious as the treatment (Eggert et al., 1995) or that suicidality resolves over time. Mediation analyses suggested that for the two treatment, increased teacher support enhanced peer support, which reduced suicidality. For all arms, perceived personal control mediated the improvements in suicidality (Thompson et al., 2000). Similarly, another study found no in suicidality between three school-based interventions: usual care (activation of social supports), Counselors-Care (C-CARE) (a 2-hour computer assisted suicide assessment followed by brief motivational counselling and activation of social supports) and Coping and Support Training (CAST) (a small group skills-training social support programme) (Eggert et al., 2002). A suicide prevention programme was also found to lead to reductions in suicidality, although the effects were comparably small (d = 0.16) (King et al., 2011). The programme involved four 50-minute psychoeducation sessions focused on depression and suicide risk factors, coping strategies and contributing factors such as family dysfunction, substance use and mental health disorders. In comparison, an RCT comparing 10 hours of psychoeducation related to psychological distress and suicide ideation to 10 hours of material focused on personal development, health and physical education classes did not find reductions between or across time (Perry et al., 2014). Discussion Based on our review of 16 interventions, overall, we found evidence that psychological treatments for depression do generally also reduce suicidality. For example, the pre to post effect of CBT on suicidality was between d = 0.34 and d = 1.02 across studies; however, a similar effect was also true of the majority of control in each of these studies. Control typically consisted of active treatment such as pharmacotherapy or TAU. The TORDIA, TADS, ADAPT, Melvin et al. (2006) and Brent et al. (1997) studies have not established superior efficacy of CBT as compared to a variety of control treatments. CBT demonstrated superior efficacy only in a single study in comparison to fluoxetine (March et al., 2009). The outcomes of this review highlight that a number of approaches show promise as interventions which address both suicidality and depression. ABFT and Interpersonal Psychotherapy ly decreased suicidal ideation in comparison to their respective control, although there were limitations in research design, attrition and sample size in these trials (Diamond et al., 2010; Tang et al., 2009). Notwithstanding such limitations, it is interesting to note that both of these promising interventions draw on relationship focused treatment approaches and this may indicate that psychological therapies that target interpersonal dynamics within personally relationships are powerful interventions for both depression and suicidality. Parent child conflict and impaired family functioning have been identified as risk factors for a suicidal event (Brent et al., 2009a; Wilkinson et al., 2011) and for depression (Lewis, 2015). Equally, family adaptability and cohesion are protective factors for adolescent suicidality (Brent et al., 2009b), so it may be expected that depression interventions, which address these critical family factors, would also demonstrate strong efficacy for suicidality. This hypothesis was supported by an RCT that found adolescents who received a family-based psychoeducation intervention for suicide behaviour experienced greater reductions in suicidal behaviour than adolescents who received routine care (Pineda and Dadds, 2013). Further analyses revealed these changes were largely

12 Devenish et al. 737 mediated by changes in family functioning (Pineda and Dadds, 2013). Another high-quality RCT, however, comparing CBT and a behaviourally focused family intervention found that neither ly reduces suicidal ideation to any greater extent than individual non-directive supportive therapy (Brent et al., 1997). Furthermore, family-based depression interventions do not always report whether they are successful in improving family functioning, reducing conflict or increasing the support adolescents receive. Thus, we do not know whether the mixed results of psychological interventions for depression on suicidality are due to their lack of impact on these family variables. There are several key areas where more research is required. One area where there is a lack of research is investigating whether specific psychological intervention approaches may be superior to other approaches in the treatment of suicidality in depressed adolescents. The larger and methodologically superior trials compare psychological treatments to medication. There is, however, a need for larger high-quality trials comparing distinctly different types of psychological therapies and a need for greater understanding of which components of a given treatment are the key therapeutic elements. For example, the secondary analyses on TADS data suggested partial completion of CBT homework may decrease suicidality; however, Simons et al. (2012) note that caution should be exercised in assuming a direct causal connection. Individuals were not randomly assigned to homework or no homework and the relationship between homework completion and suicidal ideation may be due to pre-existing between participants. Another area where studies could be improved is in terms of length of follow-up. Currently, it is unclear whether positive outcomes are sustained in the face of recurrent risk factors such as negative life events or ongoing family or peer stressors which could potentially increase suicidality. In the majority of the studies, suicidality decreased in all participants regardless of assignment to treatment group. Significant decreases in suicidality were seen in most studies, even when low levels of treatment were received. This is particularly of note in universal and selective prevention trials, where it is likely participants will enter the study with low levels of suicidal ideation, which means that identifying any changes in suicidality will be more difficult than clinical trials where adolescents enter with a high level of suicidality. The findings of a lack of changes in suicidality may also be explained by this, for example, in the universal prevention study where suicidal ideation at baseline reflected endorsement of just one item some of the time (Perry et al., 2014). The reduction in suicidality in other prevention studies therefore highlights a promising area for research. It is possible that at higher levels, suicidality is quite responsive to therapeutic attention and the activation of social support which are inherent aspects of all psychological treatments. Particularly of note was the reduction in suicidality seen in response to a brief assessment and the activation of social supports (Eggert et al., 1995, 2002). Social support is a known protective factor for suicidality in adolescents, and increasing social support for adolescents has been shown to decrease suicidality (Babiss and Gangwisch, 2009; Brausch and Decker, 2013; Gallagher et al., 2014; Ystgaard et al., 2009). However, while teacher support and peer support appeared to be important mediators of suicidality in the prevention programme, peer support, teacher support and general support were not identified as mediators of suicidality in the assessment group (Thompson et al., 2000). Personal control appeared to be another important factor, and it has been suggested that personal control and social support may supplement each other as psychological resources or protective factors (Thoits, 1995; Thompson et al., 2000). Another important line of enquiry suggested by the findings of this review is to consider the content of interventions targeting suicide only compared to interventions targeting both depression and suicide. Many interventions designed specifically to reduce suicide focus on increasing safety behaviours and help seeking in adolescents (i.e. Asarnow et al., 2015; Cotgrove et al., 1995). Depressionrelated causes of suicidal ideation and behaviour may not be a focus. Given the frequency of suicidality in depressed adolescents, it is important that future studies develop content for psychological therapies designed to address suicidality, and that depression trials routinely include adolescents with high suicidality (with the appropriate risk management in place) in order to assess the efficacy of treatment on both depression and suicidality. Possibly, future research may benefit from a more thorough examination of more subtle commonalities between suicide only interventions and those that target depression Overall, the findings of this review add a number of new dimensions to previous similar reviews. This review provides partial support for the review by Spirito et al. (2011) which found CBT depressive interventions have equivalent efficacy for suicidality when compared to other treatments. The results of our review differ to the previous review on psychological therapy for depression in adolescents by Weisz et al. (2006), which found psychological interventions have only marginal effects on suicidality. However, the review by Weisz et al. (2006) included children, whereas our review was limited to adolescents, contained several new studies, included comparison of active treatments and the focus was specifically on suicidality. In comparison to that (Weisz et al., 2006) review, our review contains a wider range of studies in that pilot studies and quasi-experimental designs were included, which may come at the expense of an increase in study heterogeneity. In addition, the current review also found some promising evidence for ABFT and Interpersonal Therapy. These approaches warrant further evaluation and replication in different populations.

13 738 ANZJP Articles Limitations There are limitations which prevent clear conclusions being made as to the efficacy of psychological depression interventions for the treatment of suicidality in adolescents. Given the high risk of bias in many of the reviewed studies, the findings of these studies should be interpreted with caution. Research to date is limited, contains methodological issues and contains smaller sample sizes than that of adult depression research populations. Comparisons across studies are also problematic due to the in measures and definition of suicidality, types and intensities of interventions, population (particularly in level of depression and suicidality) and control used. Additionally, rates of attrition and variation in retention and follow-up also varied considerably between studies, with the most robust sample sizes being found in the CBT studies, and high attrition levels leading to small sample sizes for analysis at follow-up. There is also considerable variation reported in retention during the intervention phase of studies making comparison across even CBT studies difficult. Additionally, the lack of research reporting on comparisons between psychological treatments and alternative psychological treatments ly limits conclusions regarding efficacy. Conclusion This review was driven by four research questions: Are psychological depression interventions more efficacious for the treatment of suicidality in adolescents than no treatment, placebo control and antidepressant medications, and, are there specific psychological depression treatments or approaches that have increased efficacy compared to other treatments or approaches. Psychological depression interventions show some promise for the treatment of suicidality in adolescents. In sum, the studies that have been conducted suggest psychological depression interventions are at least as efficacious as other treatment options; however, while it is unlikely that psychological treatments are no more effective than receiving no clinical treatment, there are few genuine comparisons to TAU or waiting list control. This limits our ability to rule out natural resolution of symptoms over time. To better assess whether a psychological depression treatment is a suitable treatment option for adolescents exhibiting suicidal behaviour, future research should focus on its efficacy in comparison to other psychological treatments. There is consistent evidence suggesting CBT produces moderate effects on suicidality, but at this stage, it is unclear whether these effects are sustained over the long term, similar to findings related to psychological depression intervention effects on depression in adolescents (Weisz et al., 2006). CBT is not generally more effective than pharmacological treatments for depression, and there have been mixed results as to whether combining pharmacological treatment and CBT produces stronger effects on suicidality. Further research is needed to explore whether psychological treatments that focus on reducing both depression and suicidality symptoms are more efficacious than interventions that focus on only one of these dimensions. The large effect sizes for interpersonal therapy and ABFT provide preliminary and promising evidence that relationally based interventions may be more effective than other treatments; however, direct headto-head trials are needed to establish whether these interventions are more effective than other psychological depression interventions. Additionally, there is promising evidence that depression prevention interventions may be effective in reducing suicidality in adolescents in the short term, and research evaluating prevention interventions over a longer period of time is warranted. Thus, at this point in time, there is some support for the use of psychological depression treatments to target suicidal behaviour in adolescents. They have been shown to be at least as efficacious as other treatments. However, to meet American Psychological Association (APA) clinical guidelines for treatment efficacy, further research is needed to establish whether psychological treatments provide any further benefits beyond simply being in treatment. There is a concerning lack of high-quality research assessing the efficacy of psychological intervention for suicidality in adolescents. Additionally, in depression treatments designed to improve depression and suicide symptoms through improving family functioning, there is no evidence reported as to whether family functioning is ly improved. By monitoring the effect of treatment on suicidality in depressed adolescents, future research could address the lack of clarity regarding what processes may lead to improved treatment responses in adolescents. Furthermore, relationship focused depression approaches show promise for the treatment of depression and suicidality in adolescents, and this review draws attention to the need to develop and rigorously evaluate such interventions. Acknowledgements Bethany Devenish, Lesley Berk and Andrew J Lewis contributed equally to this work Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article. Funding The author(s) received no financial support for the research, authorship and/or publication of this article. References Asarnow JR, Berk M, Hughes JL, et al. (2015) The SAFETY program: A treatment-development trial of a cognitive-behavioral family treatment for adolescent suicide attempters. Journal of Clinical Child and Adolescent Psychology 44: