Review article

Is psychotherapy effective for reducing suicide attempt and non-suicidal self-injury rates? Meta-analysis and meta-regression ofliterature data

Raffaella Calati a, b, *, Philippe Courtet a, b, c

a INSERM U1061, University of Montpellier UM1, Montpellier, Franceb FondaMental Foundation, Francec Department of Emergency Psychiatry & Acute Care, Lapeyronie Hospital, CHU Montpellier, Montpellier, France

a r t i c l e i n f o

Article history:Received 6 January 2016Received in revised form13 April 2016Accepted 14 April 2016

Keywords:PsychotherapySuicide attemptPersonality disorder

a b s t r a c t

Objective: To determine the efficacy of psychotherapy interventions for reducing suicidal attempts (SA)and non-suicidal self-injury (NSSI).Methods: Meta-analysis of randomized controlled trials (RCTs) comparing psychotherapy interventionsand treatment as usual (TAU; including also enhanced usual care, psychotropic treatment alone,cognitive remediation, short-term problem-oriented approach, supportive relationship treatment,community treatment by non-behavioral psychotherapy experts, emergency care enhanced by providereducation, no treatment) for SA/NSSI. RCTs were extracted from MEDLINE, EMBASE, PsycINFO andCochrane Library and analyzed using the Cochrane Collaboration Review Manager Software andComprehensive Meta-analysis.Results: In the 32 included RCTs, 4114 patients were randomly assigned to receive psychotherapy(n ¼ 2106) or TAU (n ¼ 2008). Patients who received psychotherapy were less likely to attempt suicideduring the follow-up. The pooled risk difference for SA was "0.08 (95% confidence intervals ¼ "0.04to "0.11). The absolute risk reduction was 6.59% (psychotherapy: 9.12%; TAU: 15.71%), yielding an esti-mated number needed to treat of 15. Sensitivity analyses showed that psychotherapy was effective for SAmainly in adults, outpatients, patients with borderline personality disorder, previously and non-previously suicidal patients (heterogeneous variable that included past history of SA, NSSI, deliberateself-harm, imminent suicidal risk or suicidal ideation), long- and short-term therapies, TAU only as acontrol condition, and mentalization-based treatment (MBT). No evidence of efficacy was found for NSSI,with the exception of MBT. Between-study heterogeneity and publication bias were detected. In thepresence of publication bias, the Duval and Tweedie's “trim and fill” method was applied.Conclusion: Psychotherapy seems to be effective for SA treatment. However, trials with lower risk of bias,more homogeneous outcome measures and longer follow-up are needed.

© 2016 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.1. Search strategy and selection criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.2. Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.3. Data extraction and quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.4. Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

* Corresponding author. INSERM U1061, University of Montpellier UM1, Mont-pellier, France.

E-mail address: raffaella.calati@gmail.com (R. Calati).

Contents lists available at ScienceDirect

Journal of Psychiatric Research

journal homepage: www.elsevier .com/locate/psychires

http://dx.doi.org/10.1016/j.jpsychires.2016.04.0030022-3956/© 2016 Elsevier Ltd. All rights reserved.

Journal of Psychiatric Research 79 (2016) 8e20

3.1. Included trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.2. Primary outcome: suicide attempt rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153.3. Secondary outcome: non-suicidal self-injury rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.4. Funnel plots and quality assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1. Introduction

In 2012, an estimated 804.000 suicide deaths (SD) occurredworldwide, corresponding to an annual suicide rate of 11.4 per100.000 people (World Health Organization, 2014). Moreover, foreach adult SD there may have been more than 20 suicide attemp-ters. Significantly, in the general population a prior suicidal attempt(SA) is the most important risk factor for suicide.

The inclusion of suicidal behavior disorder in the fifth edition ofthe Diagnostic and Statistical Manual of Mental Disorders (DSM-5)(American Psychiatric Association, 2013) highlights the need ofadditional studies to identify effective strategies for its preventionand treatment. However, one problem in research on suicide is theconfusion resulting from the use of different terms to define sui-cidal behavior (De Leo et al., 2006). Indeed, in the United States, SA(self-harm with an intention to die) is distinguished from non-suicidal self-injury (NSSI; self-harm without intention to die). Inthe United Kingdom, and more generally in Europe, and in SouthAfrica, Australia and New Zealand, deliberate self-harm (DSH;without focus on the intent) is commonly used. It includes self-harm with suicidal intent, NSSI and self-harm episodes with un-clear intent. However, DSH has been criticized because it is toobroad (Linehan, 1997). Most studies on adolescents focused on DSH(Ougrin et al., 2012; Ougrin and Boege, 2013), and DSH and NSSIprevalence in adolescents from different countries are similar(Muehlenkamp et al., 2012). Moreover, a meta-analysis of theoverall international NSSI prevalence reported that the pooled NSSIprevalence was 17.2% among adolescents, 13.4% among youngadults and 5.5% among adults (Swannell et al., 2014).

Very few evidence is available on the usefulness of specificpharmacological interventions for suicidal behavior (Hawton et al.,2015a), with the exception of the recent enthusiasm for ketamine(Bolton et al., 2015). Similarly, consensus is lacking on the effec-tiveness of psychological interventions for suicidal behavior. It hasbeen reported that a wide range of therapies, such as cognitive-behavioral therapy (CBT), dialectical behavior therapy (DBT) andproblem-solving approaches, are effective in reducing suicidalthoughts and behaviors, when they are considered as part of anextremely wide outcome variable that includes different indicators,such as SA, suicidal plans, suicidal thoughts together with hope-lessness and satisfaction with life measures (Tarrier et al., 2008).Conversely, psychosocial interventions after DSH do not seem toreduce the likelihood of subsequent SD (Crawford et al., 2007).Moreover, a recent review of randomized controlled trials (RCTs) onpsychosocial interventions for DSH in children and adolescentsfound very little evidence supporting the effectiveness of suchapproaches in these populations (Hawton et al., 2015b). This lack ofconsensus could be explained by the high between-study hetero-geneity due to the difference in suicidal phenotypes (suicidalideation, NSSI, SA, DSH, SD), treatments, diagnosis [borderlinepersonality disorder (BPD), major depressive disorder, bipolar dis-order, schizophrenia, anorexia and anxiety disorders] and

populations (adults, adolescents) included in these RCTs.A recent meta-analysis evaluated the efficacy of specific thera-

peutic (psychological, social and pharmacological) interventions inreducing any type of DSH (SA, NSSI and/or self-harm with ambig-uous intent) (19 included studies) and SA alone (8 included studies)in adolescents. Evidence of treatment efficacy was only found forthe global category of DSH, with high between-study heterogene-ity, but not for SA (Ougrin et al., 2015). Therefore, we decided toperform a new meta-analysis to extend the analysis on the efficacyof therapeutic interventions also to adults and to focus only on SAand NSSI outcomes. To this aim, in the present meta-analysis, weprimarily evaluated the efficacy of psychotherapeutic interventions(compared with treatment as usual) on the SA outcome in differentpopulations with different diagnoses. We also evaluated the effi-cacy of psychotherapeutic interventions for the treatment of NSSI/self-harming/self-mutilating behaviors (secondary outcome).Indeed, although there is a considerable overlap between SA andNSSI, the factors contributing to these two conditions could beslightly different (Dougherty et al., 2009). Finally, we also per-formed sensitivity and meta-regression analyses to take into ac-count the possible between-study heterogeneity. To our knowledgethis is the first meta-analysis that evaluated the efficacy of psy-chotherapies in specifically reducing SA and NSSI rates in bothadults and adolescents.

2. Methods

This meta-analysis was performed according the PreferredReporting Items for Systematic Reviews and Meta-Analyses State-ment (PRISMA) (Moher et al., 2009).

2.1. Search strategy and selection criteria

Records were primarily identified by a MEDLINE-based search,but results obtained by interrogating EMBASE, PsycINFO, and theCochrane Library (until June 2015) databases were also incorpo-rated. The following search terms were used: (psychotherapy ORpsychosocial OR acceptance and commitment therapy OR cognitivebehavio(u)ral therapy OR cognitive therapy OR dialectical behav-io(u)r therapy OR interpersonal psychotherapy OR mentalization-based treatment OR mindfulness based cognitive therapy ORproblem solving therapy OR schema-focused therapy ORtransference-focused psychotherapy) AND (suicid* OR self(-)harmOR non-suicidal self-injury OR self-mutilation). The reference listsof the identified studies, reviews and meta-analyses were alsoexamined to extract additional articles.

Studies were included if: they were published in a peer-reviewed journal; they were written in English; they were RCTs;they compared a form of psychotherapy (or a substantial compo-nent of psychotherapeutic methods in the treatment) with treat-ment as usual (TAU) that included also enhanced usual care (suchas a facilitated referral process with ongoing clinical monitoring),

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e20 9

psychotropic treatment alone, cognitive remediation, short-termproblem-oriented approach, supportive relationship treatment,community treatment by non-behavioral psychotherapy experts,usual emergency care enhanced by provider education, no treat-ment; they reported SA and/or NSSI as outcome measures. RCTsthat compared psychotherapy (or combined therapy) with anti-depressants were included. In this case, data on psychotherapyalone and in combinationwere pooled and compared with the dataon antidepressants. When more than two groups were present,only the psychotherapy and control groups were selected (e.g., (Weiet al., 2013)), or samples were pooled (e.g., (Brent et al., 2009)).Trials that evaluated long-term or short-term psychotherapeutictreatments were included. Short-term treatment was arbitrarilydefined as a treatment with a maximum duration of 3 months orwith a maximum of 14 sessions.

Studies were excluded if: they were performed on overlappingsamples; they focused on an intervention that did not involve pa-tients, but only their carers (e.g., Youth-Nominated Support Team(King et al., 2009)); they considered SA and NSSI/self-harming be-haviors together; they did not report SA in both groups (Rucci et al.,2011; Walkup et al., 2008; Weinberg et al., 2006) because in thiscase it was not possible to calculate the overall effect size (Sweetinget al., 2004).

2.2. Outcomes

The primary outcome was the SA rate. SA occurrence (at leastone) during the treatment and/or follow-up periodwas extracted foreach sample. When both SA and SD rates were indicated, the total(SAþ SD) rate was included (Hawton et al., 1987; Hollon et al., 1992).

The secondary outcome was the NSSI/self-harming/self-mutilating behavior rate. NSSI definition included self-injuries(most commonly, cutting or burning) without suicidal intent.Self-harming and self-mutilating behaviors were consideredtogether with NSSI onlywhen their rateswere presented separatelyfrom the SA rates. For example, in the two studies by Bateman et al.(Bateman and Fonagy,1999; 2009), the criteria for “self-mutilation”were as follows: “1) deliberate, 2) resulted in visible tissue damageand 3) nursing or medical intervention required”. These criteria aredifferent from the NSSI criteria; however, in our meta-analysis NSSIand self-mutilation were considered together because in the twomentioned studies self-mutilation acts were distinguished from SA.

When the outcomes of interest were not reported in theselected RCTs, the authors were directly contacted.

2.3. Data extraction and quality assessment

Data were screened by both the reviewers (R.C. and P.C.); dis-agreements were solved by discussion. For each RCT, the evaluatedtreatments, total sample size, percentage of females, mean age,diagnosis, assessment of medications, duration of psychothera-peutic treatment, number of weekly sessions, duration of the followup, main scales used for the assessment, and main results wereidentified (see Table 1). The quality of reporting and the risk of biasfor the included studies were assessed with the Jadad scale (Jadadet al., 1996) and also the Cochrane Risk of Bias Tool (Higgins et al.,2011). The Jadad scores were used to perform a meta-regression,while the Cochrane Risk of Bias Tool was employed because it al-lows a risk of bias evaluation of each study simply by visual in-spection of a risk of bias summary table (Figs. 2 and 3, and Figs. 1Sand 2S of the Supplementary materials). As the retained RCTs wereonly single-blind trials (due to the study type), double-blind wasnot included as a criterion. Moreover, for each study, the reportedreliability estimates (kappa and intra-class correlation coefficients -ICC) were assessed (Supplementary Table 1S).

2.4. Data analysis

Intention-to-treat (ITT) data were included when possible,otherwise completers' data were considered. To perform the meta-analyses, data were entered in and analyzed with the CochraneCollaboration Review Manager Software (RevMan, version 5.3).Individual and pooled risk differences and associated 95% confi-dence intervals (CI) were calculated. Between-study heterogeneitywas assessed with the Chi2 goodness of fit and I2 tests. The signif-icance of the pooled effect size was determined using a Z test. Datawere analyzed using a random effect framework due to theassumption of between-study heterogeneity. However, using afixed effect framework, results did not consistently change (dataavailable on request). A funnel plot was created to reveal thepreferential publication of statistically significant results. TheEgger's test was also used to evaluate the funnel plot asymmetry(Egger et al., 1997) with the STATISTICA software (Statsoft, 1995). Inthe presence of publication bias, the Duval and Tweedie's “trim andfill” method (Duval and Tweedie, 2000) was applied usingComprehensive Meta-analysis (version 2.2) to adjust the effect sizeand CIs for the missing studies and they were then assumed to beunbiased. Post hoc sensitivity and meta-regression analyses wereperformed to account for between-study heterogeneity. For thesensitivity analyses, the following subgroups of particular interest($2 RCTs) were considered: adults, adolescents, outpatients, in-patients, patients with BPD, patients with depression, patients withschizophrenia-spectrum disorders, previously and non-previouslysuicidal patients (heterogeneous variable comprising previous SA,NSSI, DSH, imminent suicidal risk or suicidal ideation), long- andshort-term psychotherapies, TAU only as a control condition, CBT,DBT, cognitive therapy (CT), mentalization-based treatment (MBT),interpersonal psychotherapy (IPT) and NSSI (strictly defined).Concerning the influence of different potential moderators (meta-regressors) on both outcomes, the following variables wereconsidered: quality of the studies (Jadad scale), gender (female) andmean age of the included patients, number of weekly sessions, totalnumber of sessions, psychotherapy duration (years) and follow-upduration (years). For meta-regression analyses, datawere evaluatedusing Comprehensive Meta-analysis (version 2.2). All p values weretwo-tailed and statistical significance was set at the 0.05 level.

3. Results

3.1. Included trials

Among the 6961 references initially retrieved, 5001 remainedafter duplicate elimination. After a first selection, 160 full-text ar-ticles were assessed for eligibility and after careful reading, 32 RCTswere retained for the analysis (Asarnow et al., 2011; Bateman andFonagy, 1999; 2009; Brent et al., 2009; Brown et al., 2005;Comtois et al., 2011; Davidson et al., 2006; Diamond et al., 2010;Donaldson et al., 2005; Esposito-Smythers et al., 2011; Guthrieet al., 2001; Hawton et al., 1987; Hollon et al., 1992; Hvid et al.,2011; Klingberg et al., 2012; Kuipers et al., 1997; Linehan et al., 1991,2006;Mcleavey et al., 1994, Mcmain et al., 2009, Morley et al., 2014;Mufson et al., 2004; Nordentoft et al., 2002; Pistorello et al., 2012;Rudd et al., 2015; Salkovskis et al., 1990; Tyrer et al., 2003; Van DerSande et al., 1997, Verheul et al., 2003; Vitiello et al., 2009; Weiet al., 2013; Wong, 2008) (see flowchart in Fig. 1 and summary inTable 1).

The retained RCTs focused on adults (n ¼ 25, 78.1%) and ado-lescents (n¼ 7, 21.9%) (studies on adolescents (Asarnow et al., 2011;Brent et al., 2009; Diamond et al., 2010; Donaldson et al., 2005;Esposito-Smythers et al., 2011; Mufson et al., 2004; Vitiello et al.,2009)). They included patients with different diagnoses: BPD

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e2010

Table 1Features of the included randomized controlled trials (RCTs) reporting the effects of psychotherapies versus treatment as usual (TAU) on suicide attempt (SA) or non-suicidal self-injury (NSSI) rates.

Study Treatments Medications Duration N of weekly sessions Followup

Sample Gender(females:N, %)

Mean age Diagnosis Mainscales

Main results Jadad

(Hawtonet al.,1987)

Outpatient counseling (OC) vsgeneral practitioner care

e Short-term treatment 8 sessions in total 1 year N ¼ 80 53, 66.25 29.34 Overdose SISRRSSASBDIGHQ

(")Two sub-groupsbenefited morefrom OC: womenand patientswith dyadicproblems

4

(Salkovskiset al.,1990)

CBT vs TAU e 1 month 5 sessions in total 1 year N ¼ 20 10, 50 27.24 Patients at high risk of repeatedSA (at least 2 previous SA)

BSSIBSISPQRSTPOMSBDIHS

CBT:Y SIY RepetitionY DepressionY Hopelessness

2

(Linehanet al.,1991)

DBT vs TAU Tapered off 1 year 2 1 year N ¼ 44 44, 100 18e45 BPD PHIBDIBHSRLI

DBT:Y ParasuicideY Medically severeparasuicide[ IndividualtherapyY Inpatientpsychiatric careduration

3

(Hollonet al.,1992)

CT (alone or combined) vsimipramine

Randomizedtreatmentassignment

3 months At most 20 sessions(14.9 ± 11.5) in total

3months

N ¼ 64 ITT:51, 80

32.6 ± 10.8 MDD BDIHRSDGAS

(") 4

(Mcleaveyet al.,1994)

Interpersonal Problem-Solving Skills Training (IPSST)vs short- term problem-oriented approach

e 3-6 months 5.3 ± 0.48 in total 6months

N ¼ 39 10, 25.6 24.44 Self-poisoning (dysthymia,dependent personalitydisorder, alcohol abuse, panicdisorder, no diagnosis)

ICPSMEPSSRPSHS

IPSST:Y Self-poisoning[ Problem solving

3

(Kuiperset al.,1997)

CBT and TAU vs TAU Assessed 9 months In the first phaseweekly

9months

N ¼ 60 22, 36.7 40.26 Medication-resistant psychosis PSEBPRSMADS

Y BPRS(")

3

(Van DerSandeet al.,1997)

Intensive in-patient andcommunity intervention vsTAU

e 1-4 days ofhospitalization andtreatment by problemsolving approach

e 12months

N ¼ 274 180, 65.7 36.29 Suicide attempters SISMADRSSCL-90HS

(") SA(") SCL-90(") HC

3

(BatemanandFonagy,1999)

Psychoanalytically orientedpartialhospitalization vs TAU

Assessed 1.45 years 1) Weekly individualpsychotherapy2) Thrice weeklygrouppsychotherapy3) Weeklyexpressive therapy4) Weeklycommunity meeting

18months

N ¼ 38 22, 57.9 31.8 BPD SCIDSSHISCL-90BDISTAISASIIP-C

Psychoanalyticallyoriented partialhospitalization:Y Suicidal and self-mutilatory actsY Inpatient caredaysY Depressivesymptoms[ Social andinterpersonalfunction

2

(Guthrieet al.,2001)

Short-term IPT vs TAU e 4 sessions in total, 1per week

6months

N ¼ 119 66, 55.46 31.2 ± 1.5 Adults who had deliberatelypoisoned themselves

SSIBDI

IPT:Y SAY SI

4

(continued on next page)

R.Calati,P.Courtet/JournalofPsychiatric

Research79

(2016)8e20

11

Table 1 (continued )

Study Treatments Medications Duration N of weekly sessions Followup

Sample Gender(females:N, %)

Mean age Diagnosis Mainscales

Main results Jadad

[ Satisfaction withtreatment

(Nordentoftet al.,2002)

Integrated treatment (IT) vsTAU

Part of the IT 2 years 1 year N ¼ 321 128, 39.9 27.0 ± 6.3 First-episode schizophrenia-spectrumdisorder

SCANEPSISIRAOS

IT:Y Hopelessness

2

(Tyrer et al.,2003)

MACT vs TAU e 3 months 7 sessions in total 1 year N ¼ 480 154, 32 32 ± 11 Patients with recurrentdeliberate self-harm

HADSGAFBHSSFQ

(") 2

(Verheulet al.,2003)

DBT vs TAU Assessed 1 year 2 1 year N ¼ 58 58, 100 34.9 ± 7.7 BPD with and withoutsubstance abuse

BPDSILPC

DBT:Y Self-mutilatingand self-damagingimpulsive behaviors[ Treatmentretention

2

(Mufsonet al.,2004)

IPT vs TAU Assessed 3 months 12 sessions in total, 1per week

3months

N ¼ 63 53, 84.13 15.12 MDD, dysthymia, depressivedisorder not otherwisespecifiedor adjustment disorder withdepressed mood

HDRSBDICGASCGISAS

IPT:Y Depressivesymptoms[ ClinicalimprovementY Clinical severity[ Globalfunctioning[ Social functioning

3

(Donaldsonet al.,2005)

Skills-based treatmentvs supportive relationshiptreatment

SSRI alone: 50%SSRI plusanothermedication:33%Atypicalantidepressant:6%Mood stabilizer:11%SBT ¼ 60%SRT ¼ 50%

6 months 10-14 sessions intotal

6months

N ¼ 39 7, 18 15.0 ± 1.7 Suicide attempters (MDD,disruptive behavior disorder,substance use disorder)

DISCSIQCES-DSTAXI

Both groups:Y SIY Depressed mood

1

(Brownet al.,2005)

CT vs enhanced usual care CT: 51.7%EUC: 53.6%

10 in total, 1, 2 or asneeded per week

18months

N ¼ 120 73, 60.8 35 MDD: 77%Substanceuse disorder: 68%

HRSDBDIBHSSSI

CT:Y SA(") SIY Self-reporteddepressionY Hopelessness

5

(Davidsonet al.,2006)

CBT and TAU vs TAU e 1 year Less than 1 (27 ± 13sessions in total)

1 year N ¼ 106 e 18e65 BPD ADSHIBDISTAIBSIIIP

CBT:Y Suicidal actsY Positive symptomdistress indexY State anxietyY Dysfunctionalbeliefs

5

(Linehanet al.,2006)

DBT vs communitytreatment by nonebehavioralpsychotherapy experts

Assessed 1 year 2 2 years N ¼ 101 101, 100 29.29 BPD with recent suicidal andself-injurious behaviors (atleast 2 SA or self-injuries in thepast 5 years)

SASIISBQRLITHIHRSD

DBT:Y SAY Hospitalizationfor SIY Medical risk

5

R.Calati,P.Courtet/JournalofPsychiatric

Research79

(2016)8e20

12

Y Drop outY PsychiatrichospitalizationsY Psychiatricemergencydepartment visits

(Wong,2008)

CBT vs control group (notreatment)

Assessed 10 weeks 10 sessions in total 10weeks

N ¼ 96 80, 79 37.4 ± 9.4 MDD BDIECDAS

CBT:Y Depression[ Coping skillsY NegativeemotionsY Dysfunctionalattitudes

2

(BatemanandFonagy,2009)

MBT vs structured clinicalmanagement

Assessed 18 months 140 sessions in total,2 per week

18months

N ¼ 134 107, 79.8 31.11 BPD GAFSCL-90BDISASIIP-C

MBT:Y SAY Hospitalization

5

(Brent et al.,2009)

CBT and SSRI or venlafaxine vsSSRI or venlafaxine alone

Randomizedassignment totreatment

3 months Up to 1 per week(8.3 ± 3.6)

3months

N ¼ 334 e 12e18 MDDnot responders to a previousSSRI

K-SADS-PLCGICGASCDRSBDIBHSSIQ

(") 3

(Mcmainet al.,2009)

DBT vs general psychiatricmanagement

81.6% 1 year 2 2 years N ¼ 180 155, 86.1 30.4 ± 9.9 BPD with at least 2 suicidal orNSSI episodes in the past 5years

SASIIZRSSCL-90-RSTAXIBDIIPP-C

No differencebetween groups

3

(Vitielloet al.,2009)

We grouped CBT (alone orcombination) vs fluoxetine orplacebo

Randomizedassignment totreatment

36 weeks 24-30 sessions intotal

36weeks

N ¼ 439 237, 54 14.6 ± 1.5 MDD C-CASASIQCDRSADSCGIBHSRADS

CBT:Y SA

3

(Diamondet al.,2010)

Attachment-Based FamilyTherapy (ABFT) vs enhancedusual care

Assessed Around 10 in total 6months

N ¼ 66 55, 83.3 15.19 MDD, anxiety disorder orexternalizing disorders (ADHD,ODD, CD)

SSIBDI

ABFT:Y SI

3

(Asarnowet al.,2011)

Family-based CBT vs usualemergency department careenhanced byprovider education

Assessed 1 month 1 session in total 2months

N ¼ 181 125, 69 14.7 ± 2.0 Suicidal youths (SA and/or SI) SACAHASSCES-DCBCLCBQ

CBT:[ OutpatienttreatmentBoth:(") clinical/functioningoutcomes

5

(Comtoiset al.,2011)

CAMS vs enhanced usual care Assessed 1-3 months 4-12 weekly sessionsin total

1 year N ¼ 32 14, 38 36.8 ± 10.1 Suicidal patients (recent SA orimminent suicidal risk)

CSQSSIOQ-45RLIOHSSASII

CAMS:Y SIY Symptom distress[ Hope

4

(Esposito-Smytherset al.,2011)

CBT vs enhanced TAU Assessed 1 year Weekly (6 months),bi-monthly (3months) andmonthly (3 months)

18months

N ¼ 40 24, 66.7 15.72 ± 1.19 Co-occurring alcohol or otherdrug use disorder andsuicidality (SA or SI)

K-SADSSIQCIS

CBT:Y SAY Globalimpairment

3

(continued on next page)

R.Calati,P.Courtet/JournalofPsychiatric

Research79

(2016)8e20

13

Table 1 (continued )

Study Treatments Medications Duration N of weekly sessions Followup

Sample Gender(females:N, %)

Mean age Diagnosis Mainscales

Main results Jadad

Y Inpatientpsychiatrichospitalizations,emergencydepartmentvisits, and arrests

(Hvid et al.,2011)

OPAC program (outreach,problem solving,adherence, continuity) vs TAU

e 6 months 1 year N ¼ 133 95, 71.4 37.06 Wide range of diagnoses SuicidalactsRepetition

Y SAY Repetitive acts

4

(Klingberget al.,2012)

CBT vs cognitive remediation Assessed 9 months 20 sessions in total 1 year N ¼ 198 87, 43.9 36.9 ± 9.9 SKZ Severeadverseevents

No differencebetween groups

4

(Pistorelloet al.,2012)

DBT vs optimized TAU DBT: 41.9%TAU: 37.5%

1 year 2 18months

N ¼ 63 51, 81 20.86 ± 1.92 College students with at least 1lifetime NSSI act or SA and 3 ormore BPD diagnostic criteria

SCID-ISCID-IISASIIBDISBQSASGAF

DBT:Y SuicidalityY NSSIeventsY DepressionY BPD criteriaY Psychotropicmedication use[ Social adjustment

4

(Wei et al.,2013)

CT vs control group (notreatment) vs telephoneintervention

Assessed 3 months 10 sessions in total 1 year N ¼ 159 119,74.84

31.75 Patients who had attemptedsuicide

BSSIHRSD

(") 3

(Morleyet al.,2014)

Opportunistic CBT vs TAU e 6 months 8 sessions in totalplus group therapy

6months

N ¼ 185 68, 36.76 36.09 Substance use disorders SSIHADSSES

(") 3

(Rudd et al.,2015)

CBT vs TAU Assessed 3 months 12 sessions in total 2 years N ¼ 152 133, 87.5 27.40 Military personnel (MDD, PTSD,substance dependence, otheranxiety disorders)

SCID-ISCID-IISASIIBSSIBDIBAIBHS

CBT:Y SA

4

[: increase; Y: decrease; ("): no difference; SA: suicide attempt; NSSI: non-suicidal self-injury; SI: suicidal ideation; CAMS: collaborative assessment and management of suicidality; CBT: cognitive behavioral therapy; CT:cognitive therapy; DBT: dialectical behavior therapy; IPT: interpersonal psychotherapy; MACT: manual-assisted cognitive behavior therapy; MBT: mentalization-based treatment; TAU: treatment as usual; BPD: borderlinepersonality disorder; MDD: major depressive disorder; PTSD: post-traumatic stress disorder; SKZ: schizophrenia; ADS: Adolescent Depression Scale; ADSHI: Acts of Deliberate Self-Harm Inventory; BAI: Beck Anxiety Inventory;BDI: Beck Depression Inventory; BHS: Beck Hopelessness Scale; BPDSI: Borderline Personality Disorder Severity Index; BPRS: Brief Psychiatric Rating Scale; BSI: Brief Symptom Inventory; BSIS: Beck Suicidal Intent Scale; BSSI:Beck Scale for Suicide Ideation; CBCL: Child Behavior Checklist; CBQ: Conflict Behavior Questionnaire; C-CASA: Columbia Classification Algorithm of Suicidal Assessment; CDRS: Child Depression Rating Scale; CES-D: Center forEpidemiological Studies Depression Scale; CGAS: Children's Global Assessment Scale; CGI: Clinical Global Impressions scale; CIS: Columbia Impairment Scale; CSQ: Client Satisfaction Questionnaire; DAS: Dysfunctional AttitudeScale; DISC: Diagnostic Interview Schedule for Children; EC: Emotions Checklist; EPSIS: European Parasuicide Study Interview Schedule; GAF: Global Assessment of Functioning; GAS: Global Assessment Scale; GHQ: GeneralHealth Questionnaire; HADS: Hospital Anxiety and Depression Scale; HASS: Youth report on the Harkavy Hasnis Scale; HRSD: Hamilton Rating Scale for Depression; HS: Hopelessness Scale; ICPS: Interpersonal Cognitive ProblemSolving; IFR: Index of Family Relations; IIP-C: Inventory of Interpersonal Problems - Circumflex version; IRAOS: Interview for Retrospective Assessment of Onset of Schizophrenia; K-SADS-PL: School Age Schedule for AffectiveDisorders and Schizophrenia for School-Aged Children-Present and Lifetime Version; LPC: Lifetime Parasuicide Count; MADRS: Montgomery - Åsberg Depression Rating Scale; MADS: Maudsley Assessment of DelusionsSchedule; MEPS: Means-Ends Problem-Solving; OHS: Optimism and Hope Scale; OQ-45: Outcome Questionnaire-45; PHI: Parasuicide History Interview; POMS: Profile of Mood States; PQRST: Personal Questionnaire RapidScaling Technique; PSE: Present Status Examination ICD-10; RADS: Reynolds Adolescent Depression Scale; RLI: Reasons for Living Inventory; RRS: Risk of Repetition Scale; SACA: Service Assessment for Children and Adolescents;SAS: Social Adjustment Scale; SASII: Suicide Attempt Self-Injury Interview; SBQ: Suicidal Behaviors Questionnaire; SCAN: Schedules for Clinical Assessment in Neuropsychiatry; SCID-I and eII: Structured Clinical Interview forDSM; SCL-90: Symptom Checklist-90; SES: Self-Efficacy Scale; SFQ: Social Functioning Questionnaire; SIQ: Suicidal Ideation Questionnaire; SIS: Suicide Intent Scale; SRPS: Self-Rating Problem-Solving scale; SSBS: Spectrum ofSuicidal Behavior Scale; SSHI: Suicide and Self-Harm Inventory; SSI: Scale for Suicide Ideation; STAI: State-Trait Anxiety Inventory; STAXI: Spielberger Anger Expression Inventory; THI: Treatment History Interview; ZRS:Zanarini Rating Scale.

R.Calati,P.Courtet/JournalofPsychiatric

Research79

(2016)8e20

14

(n ¼ 8, 25%) (Bateman and Fonagy, 1999; 2009; Davidson et al.,2006; Linehan et al., 1991, 2006; Mcmain et al., 2009, Pistorelloet al., 2012; Verheul et al., 2003), depression (n ¼ 6, 18.7%) (Brentet al., 2009; Brown et al., 2005; Hollon et al., 1992; Mufson et al.,2004; Vitiello et al., 2009; Wong, 2008), schizophrenia-spectrumdisorders (n ¼ 3, 9.4%) (Klingberg et al., 2012; Kuipers et al., 1997;Nordentoft et al., 2002), and previous “suicidality” (past history ofSA, NSSI, DSH, imminent suicidal risk or suicidal ideation) (n ¼ 14,43.7%) (Asarnow et al., 2011; Comtois et al., 2011; Donaldson et al.,2005; Esposito-Smythers et al., 2011; Guthrie et al., 2001; Hawtonet al., 1987; Linehan et al., 2006; Mcleavey et al., 1994, Mcmainet al., 2009, Pistorello et al., 2012; Salkovskis et al., 1990; Tyrer et al.,2003; Van Der Sande et al., 1997, Wei et al., 2013). These RCTsevaluated the effects of long-term (n ¼ 15, 46.9%) and short-termpsychotherapies (n ¼ 17, 53.1%) (trials on short-term psychother-apies: (Asarnow et al., 2011; Brent et al., 2009; Brown et al., 2005;Comtois et al., 2011; Diamond et al., 2010; Donaldson et al., 2005;Guthrie et al., 2001; Hawton et al., 1987; Hollon et al., 1992;Mcleavey et al., 1994, Mufson et al., 2004; Rudd et al., 2015;Salkovskis et al., 1990; Tyrer et al., 2003; Van Der Sande et al.,1997, Wei et al., 2013; Wong, 2008). The most represented psy-chotherapeutic treatments were: CBT (n ¼ 13, 40.6%) (Asarnowet al., 2011; Brent et al., 2009; Davidson et al., 2006; Donaldsonet al., 2005; Esposito-Smythers et al., 2011; Klingberg et al., 2012;Kuipers et al., 1997; Morley et al., 2014; Rudd et al., 2015; Salkovskiset al., 1990; Tyrer et al., 2003; Vitiello et al., 2009;Wong, 2008), DBT(n ¼ 5, 15.6%) (Linehan et al., 1991, 2006; Mcmain et al., 2009;Pistorello et al., 2012; Verheul et al., 2003), CT (n ¼ 3, 9.4%)(Brown et al., 2005; Hollon et al., 1992; Wei et al., 2013),

psychoanalytically-oriented partial hospitalization or MBT (n ¼ 2,6.2%) (Bateman and Fonagy, 1999; 2009) and IPT (n ¼ 2, 6.2%)(Guthrie et al., 2001; Mufson et al., 2004). In total, 4114 patientswere randomly assigned to receive a psychotherapeutic treatment(n ¼ 2106) or TAU (n ¼ 2008).

3.2. Primary outcome: suicide attempt rate

Thirty-one trials reported the occurrence of one or more SA(Fig. 2) (Asarnow et al., 2011; Bateman and Fonagy, 1999; 2009;Brown et al., 2005; Comtois et al., 2011; Davidson et al., 2006;Diamond et al., 2010; Donaldson et al., 2005; Esposito-Smytherset al., 2011; Guthrie et al., 2001; Hawton et al., 1987; Hollon et al.,1992; Hvid et al., 2011; Klingberg et al., 2012; Kuipers et al., 1997;Linehan et al.,1991, 2006;Mcleavey et al.,1994, Mcmain et al., 2009,Morley et al., 2014; Mufson et al., 2004; Nordentoft et al., 2002;Pistorello et al., 2012; Rudd et al., 2015; Salkovskis et al., 1990; Tyreret al., 2003; Van Der Sande et al., 1997, Verheul et al., 2003; Vitielloet al., 2009; Wei et al., 2013; Wong, 2008). Patients allocated toreceive psychotherapy were less likely to attempt suicide duringthe follow-up period in comparison with patients in the TAU arm(z ¼ 4.12, p < 0.0001). The pooled risk difference for SA was "0.08(95% CI ¼ -0.04 to"0.11). The pooled difference between the risk ofSA in the psychotherapy arm (9.12%) and in the TAU arm (15.71%)gave an absolute risk reduction of 6.59%, corresponding to anumber needed to treat of 15 to prevent one SA. As high between-study heterogeneity was detected (c2 ¼ 125.82, d.f. ¼ 30,p < 0.00001, I2 ¼ 76%), post hoc sensitivity analyses were per-formed in subgroups of particular interest ($2 RCTs) (Table 2).

Records iden fied through database searching

(n = 6961)

Screening

Included

Eligibility

IdenƟfi

caƟo

n

Addi onal records iden fied through other sources

(n = 25)

Records a er duplicates removed(n = 5026)

Records screened(n = 500)

Records excluded(n = 4526)

Full-text ar cles assessed for eligibility

(n = 160)

Full-text ar cles excluded(n = 340)Studies:

- not per nent (n=266)- reviews or meta-analyses

(n=31);- did not report suicide a empt/NSSI rates for

each group (n=37);- compared different psychotherapeu c

interven ons (n=6).

Studies included in quan ta ve synthesis

(meta-analysis)(n = 32)

Full-text ar cles excluded through exclusion criteria

(n = 128)

Fig. 1. PRISMA flow diagram.

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e20 15

Fig. 2. Forest plot and risk of bias evaluation of trials comparing the effect of psychotherapeutic treatments and treatment as usual (TAU) on suicide attempt rates.

Table 2Results of the sensitivity analyses (NSSI: non-suicidal self-injury; BPD: borderline personality disorder; CBT: cognitive behavioral therapy; DBT: dialectical behavior therapy;CT: cognitive therapy; MBT: mentalization-based treatment or psychoanalytically oriented partial hospitalization; IPT: interpersonal psychotherapy).

z p c2 d.f. p I2 (%)

Suicide attemptsAdults 4.07 <0.0001 119.71 24 <0.00001 80Adolescents 0.73 0.47 9.27 5 0.10 46Outpatients 3.88 0.0001 107.60 27 <0.00001 75Inpatients 1.39 0.17 22.22 2 <0.0001 91Completers 1.81 0.07 24.91 5 0.0001 80BPD 2.85 0.004 39.75 7 <0.00001 82Depressed patients 1.37 0.17 12.69 4 0.01 68Schizophrenia-spectrum 0.42 0.68 0.88 2 0.64 0Previously suicidal patients 2.34 0.02 26.88 13 0.01 52Non-previously suicidal patients 3.31 0.0009 110.76 16 <0.00001 86Long-term psychotherapies 3.14 0.002 101.00 14 <0.00001 86Short-term psychotherapies 2.84 0.005 28.64 15 0.02 48TAU only as control condition 2.83 0.005 71.95 16 <0.00001 78CBT 1.56 0.12 28.04 11 0.003 61DBT 1.71 0.09 17.35 4 0.002 77CT 1.54 0.12 8.14 2 0.02 75MBT 2.19 0.03 6.89 1 0.009 85IPT 0.53 0.60 13.56 1 0.0002 93

Non-suicidal self-injuryAdults 1.28 0.20 18.40 6 0.005 67Outpatients 0.26 0.80 14.46 6 0.02 58BPD 1.08 0.28 18.40 5 0.002 73Previously suicidal patients 0.19 0.85 2.58 2 0.28 22Non previously suicidal patients 1.15 0.25 24.01 4 <0.0001 83Long-term psychotherapies 1.08 0.28 18.40 5 0.002 73Short-term psychotherapies 0.01 0.99 5.47 1 0.02 82TAU only as control condition 0.82 0.41 15.17 5 0.01 67CBT 0.39 0.69 5.61 2 0.06 64DBT 0.40 0.69 1.83 2 0.40 0MBT 2.19 0.03 3.17 1 0.08 68NSSI (strictly defined) 1.51 0.13 1.50 2 0.47 0

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e2016

Significant results were found for: adults (p < 0.0001); outpatients(p ¼ 0.0001); patients with BPD (p ¼ 0.004); previously suicidalpatients (p ¼ 0.02) (i.e., with a past history of SA, NSSI, DSH,imminent suicidal risk or suicidal ideation); non-previously sui-cidal patients (p ¼ 0.0009) (i.e., without a past history of SA, NSSI,DSH, imminent suicidal risk or suicidal ideation); long-term psy-chotherapies (p ¼ 0.002); short-term psychotherapies (p ¼ 0.005);studies with TAU only as control condition (p ¼ 0.005) (i.e., studieswhere TAUwas the only control treatment without any of the othernon-psychotherapy treatments we pooled in the primary analysis);psychoanalytically-oriented partial hospitalization or MBT(p ¼ 0.03).

In the meta-regression analyses, the Jadad score (p ¼ 0.01),gender (p ¼ 0.007) and number of weekly sessions (p ¼ 0.001)showed a plausible, moderating effect on the outcome (SA rate)(Table 3). Particularly, high Jadad score, female gender and highnumber of psychotherapy sessions per week were associated withsignificant SA risk reduction.

3.3. Secondary outcome: non-suicidal self-injury rate

Eight trials reported the occurrence of NSSI/self-harming be-haviors (Fig. 3) (Bateman and Fonagy, 1999; 2009; Brent et al.,2009; Davidson et al., 2006; Mcmain et al., 2009, Pistorello et al.,2012; Tyrer et al., 2003; Verheul et al., 2003). No difference in theNSSI/self-harming behavior rate was found between patients in thepsychotherapy arm and those in the TAU arm during the follow-upperiod (z ¼ 0.98, p ¼ 0.33), with a high between-study heteroge-neity (c2 ¼ 26.44, d.f. ¼ 7, p ¼ 0.0004, I2 ¼ 74%). In post hocsensitivity analyses, significant results were found only for psy-choanalytically oriented partial hospitalization and MBT (p ¼ 0.03).

3.4. Funnel plots and quality assessment

The funnel plots for the studies included in the two main ana-lyses are shown in Figs. 3S and 4S, respectively (Supplementarymaterial). Visual inspection highlighted funnel plot asymmetry forthe studies focused on SA outcome and the Egger's test was sig-nificant (p ¼ 0.001), indicating the presence of publication bias. Anadditional analysis to adjust for publication bias using the “trim andfill” method showed that the adjusted pooled risk difference for SAwas "0.07 (95% CI ¼ "0.04 to "0.10). Conversely, the Egger's testwas not significant (p ¼ 0.17) for studies on NSSI/self-harmingbehavior outcome, with no evidence of publication bias.

The risk of bias graph (Fig. 1S) and the risk of bias summary(Fig. 2S) as well as the Jadad total scores (Table 1) summarize theresults of the analyses to highlight other bias types. Specifically,there was a high rate of unclear risk of bias because of the lack ofinformation in the studies. Blinding was associated with the high-est risk of bias, due to the specific design of the retained RCTs.Conversely, sequence generation showed the lowest risk of bias.Concerning the Jadad scale (quality of reporting), five (15.6%)studies obtained the maximum score (5), eight (25%) received ascore of 4, and twelve (37.5%) a score of 3, indicating that most ofthe included studies were of good quality (78.1%). Six (18.7%)studies received a score of 2, and only one (3.2%) obtained 1.

For each trial the reliability estimates are shown inSupplementary Table 1S. Kappa was reported in ten studies (31.2%)and ICC in seven studies (21.9%). Overall, reliability was good(k ¼ 0.72e0.80) and very good (k ¼ 0.81e1.00 or ICC ¼ 0.83e1.00).

4. Discussion

The primary aim of this meta-analysis was to evaluate the

Table 3Results of the meta-regression analyses of potential moderators of the suicide attempt outcome.

Estimate SE Lower limit Upper limit z p

Suicide attemptsJadad "0.25 0.10 "0.44 0.06 "2.59 0.01Gender (female) "1.40 0.52 "2.42 "0.38 "2.68 0.007Age "0.01 0.02 "0.04 0.03 "0.40 0.69Weekly sessions "0.52 0.16 "0.84 "0.20 "3.18 0.001Total sessions "0.01 0.01 "0.01 0.01 "1.49 0.14Psychotherapy duration (years) 0.16 0.22 "0.27 0.59 0.73 0.46Follow-up duration (years) "0.32 0.23 "0.78 0.13 "1.39 0.16

Fig. 3. Forest plot and risk of bias evaluation of trials comparing the effect of psychotherapeutic treatments and treatment as usual (TAU) on non-suicidal self-injury (NSSI)/self-harming behavior rates.

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e20 17

efficacy of psychotherapeutic treatments on SA rate in adults andadolescents. Patients allocated to receive psychotherapy (31 of theincluded studies) were less likely to attempt suicide during thefollow-up period in comparison with patients allocated to receiveTAU or a similar condition. This finding is in agreement with theresults of a Danish matched cohort study showing that the risk ofrepeated DSH and mortality was lower in patients who receivedpsychosocial therapy after DSH compared with people who did not(Erlangsen et al., 2015). The present result is encouraging; however,the between-study heterogeneity was important, consistently withthe high number of heterogeneous variables: different populations(adults and adolescents), diagnoses, psychotherapeutic in-terventions and treatment duration. To account for this high het-erogeneity, we performed a number of post hoc sensitivity andmeta-regression analyses. Sensitivity analyses highlighted someclinically meaningful results. First, psychotherapy was effective inadults, but not in adolescents. This is consistent with very recentfindings on SA in adolescents (Ougrin et al., 2015) (when SAoutcome has been considered separately from DSH as a globalcategory) and also in adults (Tarrier et al., 2008). Similarly, theanalysis by Hawton and colleagues of 11 RCTs on psychosocial in-terventions for DSH in children and adolescents (Hawton et al.,2015b) showed the paucity of evidence for effective interventions.These authors followed an extremely rigorous methodology byconsidering only studies on children and adolescents and byavoiding pooling them in most of the analyses. We used anotherapproach to give a very broad overview of all the studies on theefficacy of psychotherapeutic treatments for SA in both adults andadolescents. Moreover, not all the studies retained in the two pre-viousmeta-analyses (Ougrin et al., 2015; Hawton et al., 2015b) couldbe included here because of different inclusion/exclusion criteria.

Our analysis then shows that psychotherapy was effective inoutpatients, but not in inpatients. These results could be related tothe severity of inpatients’ symptomatology. On the other hand,inpatients could have better outcomes because they can receivemore intensive treatments.

In addition, the efficacy of psychotherapy was not detected incompleters. This result could be explained by the small number ofstudies included in the sensitivity analysis (k ¼ 6). Moreover, atendency toward association (p ¼ 0.07) should be underlined.

Our analysis also shows that psychotherapy was effective inpatients with BPD, but not in patients with depression orschizophrenia-spectrum disorders. These results suggest that SAtreatment in these specific populations (patients with depressionor schizophrenia) requires the improvement of existing clinicalstrategies. Similarly, psychotherapy was effective in patients withand also in thosewithout past history of “suicidality”, considered asa heterogeneous variable comprising previous SA, NSSI, DSH,imminent suicidal risk or suicidal ideation. This is in contrast with aprevious meta-analysis that reported the lack of psychotherapyefficacy in preventing SD in patients with a past DSH episode(Crawford et al., 2007). However, it should be stressed that wefocused on SA, and not on SD, and previous suicidality is not thesame as previous DSH.

Furthermore, our study found that both long- and short-termpsychotherapies were effective, with a slightly higher effect sizefor long-term psychotherapies, and a lower heterogeneity forshort-term treatments. Finally, psychoanalytically oriented partialhospitalization or MBT was the only effective treatment. However,only two small-sized studies were included. Moreover, one of thesestudies focused on partial hospitalization treatment and this couldexplain the better SA outcome. Indeed, partial hospitalizationtreatments are more intensive and also the suicide risk duringtreatment may be reduced because of the in-hospital treatmentsetting. Meta-regression analyses led to interesting findings as well,

showing that intensive psychotherapy (high number of sessions perweek) was associated with SA risk reduction. Therefore, psycho-therapeutic approaches, especially intensive treatments, could beeffective.

Unfortunately, studies on other forms of psychotherapy (e.g.,transference-focused psychotherapy (Clarkin et al., 2001) andacceptance and commitment therapy (Ducasse et al., 2014)) couldnot be included in the main analyses and it would be interesting toevaluate them in the future.

The presence of publication bias related to the studies con-cerning the primary outcome slightly moderates the enthusiasmfor these results. We tried to correct this bias by imputing thepresence of missing studies to obtain an unbiased estimate.Nevertheless, this point highlights the need of additional studies.Furthermore, the benefits of psychotherapy were much lessimportant in large, independently funded trials than in smallerstudies and for this reason the present results should be consideredwith further caution. Independent replications of the initial positiveRCTs also could increase the confidence in the reported results.New studies should be performed not only by the people whodeveloped a specific intervention, as in the case of MBT, but also byindependent investigators. Our secondary aim was to evaluate theefficacy of psychotherapeutic treatments for NSSI/self-harming/self-mutilating behavior. For this analysis, we included onlystudies (n ¼ 8) that specifically distinguished self-harming/self-mutilating behaviors from SA, with the aim of segregating a high-ly homogeneous, phenomenologically distinct clinical entity. Wefound no difference in the rate of NSSI/self-harming behaviorsbetween patients allocated to receive psychotherapy and thosewho received TAU, with high between-study heterogeneity. This isconsistent with recent meta-analytical findings on adolescents(Ougrin et al., 2015), although qualitatively contrasting results havebeen reported (Turner et al., 2014). As most of the included studiesfocused on adults (only one RCT concerned adolescents), this resultseems to be stable across different populations. Sensitivity analysesreported the specific efficacy only of psychoanalytically orientedpartial hospitalization orMBT, consistently with previous results onadults.

As underlined by Ougrin et al. in the case of adolescents (Ougrinet al., 2015), this reported lack of efficacy may suggest that psy-chotherapeutic strategies targeting NSSI should focus on differentaspects then those targeting SA. Actually this lack of efficacy, withthe exception of MBT, could be linked to the specific personalityprofile of NSSI patients, that make them less likely to respond topsychotherapeutic approaches. Moreover, the fact that in ouranalysis patients with a history of SA and NSSI and patients withNSSI alone were probably pooled should be taken into accounttogether with the reported differences between these two groups(Perez et al., 2014). However, this represents a constant challengebecause NSSI is a strong predictor of a future SA (Victor andKlonsky, 2014).

Considering that the quality of the included RCTs might haveinfluenced the present findings, the design of future studies shouldbe substantially improved. As suggested by Bateman and Fonagyconcerning the assessment of a treatment for BPD (Bateman andFonagy, 2009), “a randomized design for assessing such a treat-ment must meet the following minimal criteria: 1) a comparisongroup also receiving a manualized, structured treatment withequivalent supervision; 2) delivery of both by professionals trainedto similar levels; 3) statistical power to detect relatively small dif-ferences; and 4) a representative sample of clinically referred pa-tients with a confirmed diagnosis […] at high risk of suicide”.

Moreover, additional aspects should be considered: variableslinked to the therapist (the use of specific therapy manuals and/orsupervision groups; duration of training and supervision; years of

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e2018

post-training experience; specific training on suicide treatment)and to the psychotherapy [treatments that specifically target sui-cidal phenotypes or not; association with psychotropic drugs(pharmacological classes and dosages), concurrently or sequen-tially; concomitant use of other therapeutic treatments (self-helpgroups, family or couple therapy, psycho-educational groups)].

Themain strong points of this meta-analysis are its large samplesize (4114 patients) and the fact that it is, to our knowledge, the firstevaluation of the effectiveness of psychotherapies for reducingspecifically SA and NSSI rates in both adults and adolescents.However, a major limitation is that extremely different studieswere pooled together. Therefore, we suggest that the reported re-sults should be considered as a starting point for further method-ologically accurate investigations. Moreover, we decided not toperform the Bonferroni correction for all the sensitivity analysesand we did not exclude the outliers (results available on request).Indeed, after the Bonferroni correction (p % 0.0016), only someresults on SA would have remained: the efficacy on adults, out-patients and non-previous suicidal patients. Moreover, after theexclusion of outliers (including standardized residuals), we couldnot have shown the results on MBT because the two studies byBateman et al. were outliers (Bateman and Fonagy, 1999; 2009). Wefollowed this approach to preserve results that, in our opinion,were clinically interesting.

A number of further limitations should be listed: 1) thebetween-study heterogeneity and the presence of publication bias;2) the small sample size of the NSSI meta-analysis. This couldrepresent a possible explanation for the reported lack of efficacy ofpsychotherapeutic treatments in NSSI. Moreover, considering thelow number of included trials (n < 10) concerning NSSI, the pres-ence of a publication bias cannot be totally excluded; 3) publicationbias is only one of the possible causes of funnel plot asymmetry forthe study on SA; 4) most of the included studies concerned smallpatient populations (e.g., (Salkovskis et al., 1990; Donaldson et al.,2005; Esposito-Smythers et al., 2011; Comtois et al., 2011)); 5)several of the included studies focused on treatments that do notspecifically target SA or NSSI (e.g., (Hollon et al., 1992; Wong, 2008;Kuipers et al., 1997)); 6) not all included studies considered SA orNSSI as primary outcomes (e.g., (Wong, 2008; Mufson et al., 2004));7) the inclusion of studies with randomized antidepressantassignment or other forms of psychotherapy in the TAU arm (e.g.,(Brent et al., 2009; Donaldson et al., 2005)); 8) the between-studyheterogeneity in the follow-up duration could have led to exposurevariability (however, we included this factor as a moderator in themeta-regression); 9) the methodological approach of excludingstudies with no events or no reported events may have led to abiased sample; however, only three studies were excluded for thisreason; 10) the lack of consideration of not published articles couldhave led to the risk of missing failed trials; 11) articles not writtenin English and that could have brought additional insights intopsychotherapeutic treatments in suicide were not included; 12) notall the potentially eligible studies were retained because of the lackof access to such articles, lack of data related to the outcomes ofinterest and lack of replies from the contacted authors.

In conclusion, psychotherapy seems to be effective for SAtreatment. However, trials with lower risk of bias, more homoge-neous outcome measures and longer follow-up are needed toconfirm the findings of this meta-analysis.

Acknowledgments

We are grateful to the authors who replied to our request foradditional study data: Cedar R. Koons, Shelley McMain and RyanBarnhart, Kirsten Morley, and Igor Weinberg. We would like tothank Dr. Elisabetta Andermarcher for the careful linguistic revision

of the manuscript.Dr. Raffaella Calati received a grant from FondaMental Founda-

tion, France. Prof. Philippe Courtet received research grants fromServier, and fees for presentations at congresses or participation inscientific boards from AstraZeneca, Bristol-Myers-Squibb, Janssen,Lilly, Lundbeck, Otsuka, Roche and Servier.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jpsychires.2016.04.003.

References

American Psychiatric Association, 2013. Diagnostic and Statistical Manual of MentalDisorders, fifth ed. American Psychiatric Association, Washington DC.

Asarnow, J.R., Baraff, L.J., Berk, M., Grob, C.S., Devich-Navarro, M., Suddath, R.,Piacentini, J.C., Rotheram-Borus, M.J., Cohen, D., Tang, L., 2011. An emergencydepartment intervention for linking pediatric suicidal patients to follow-upmental health treatment. Psychiatr. Serv. 62, 1303e1309.

Bateman, A., Fonagy, P., 1999. Effectiveness of partial hospitalization in the treat-ment of borderline personality disorder: a randomized controlled trial. Am. J.Psychiatry 156, 1563e1569.

Bateman, A., Fonagy, P., 2009. Randomized controlled trial of outpatientmentalization-based treatment versus structured clinical management forborderline personality disorder. Am. J. Psychiatry 166, 1355e1364.

Bolton, J.M., Gunnell, D., Turecki, G., 2015. Suicide risk assessment and interventionin people with mental illness. BMJ 351, h4978.

Brent, D.A., Emslie, G.J., Clarke, G.N., Asarnow, J., Spirito, A., Ritz, L., Vitiello, B.,Iyengar, S., Birmaher, B., Ryan, N.D., Zelazny, J., Onorato, M., Kennard, B.,Mayes, T.L., Debar, L.L., Mccracken, J.T., Strober, M., Suddath, R., Leonard, H.,Porta, G., Keller, M.B., 2009. Predictors of spontaneous and systematicallyassessed suicidal adverse events in the treatment of SSRI-resistant depressionin adolescents (TORDIA) study. Am. J. Psychiatry 166, 418e426.

Brown, G.K., Ten Have, T., Henriques, G.R., Xie, S.X., Hollander, J.E., Beck, A.T., 2005.Cognitive therapy for the prevention of suicide attempts: a randomizedcontrolled trial. JAMA 294, 563e570.

Clarkin, J.F., Foelsch, P.A., Levy, K.N., Hull, J.W., Delaney, J.C., Kernberg, O.F., 2001. Thedevelopment of a psychodynamic treatment for patients with borderline per-sonality disorder: a preliminary study of behavioral change. J. Pers. Disord. 15,487e495.

Comtois, K.A., Jobes, D.A., S. O'Connor, S., Atkins, D.C., Janis, K., Chessen, C.,Landes, S.J., Holen, A., Yuodelis-Flores, C., 2011. Collaborative assessment andmanagement of suicidality (CAMS): feasibility trial for next-day appointmentservices. Depress Anxiety 28, 963e972.

Crawford, M.J., Thomas, O., Khan, N., Kulinskaya, E., 2007. Psychosocial in-terventions following self-harm: systematic review of their efficacy in pre-venting suicide. Br. J. Psychiatry 190, 11e17.

Davidson, K., Norrie, J., Tyrer, P., Gumley, A., Tata, P., Murray, H., Palmer, S., 2006. Theeffectiveness of cognitive behavior therapy for borderline personality disorder:results from the borderline personality disorder study of cognitive therapy(BOSCOT) trial. J. Pers. Disord. 20, 450e465.

De Leo, D., Burgis, S., Bertolote, J.M., Kerkhof, A.J., Bille-Brahe, U., 2006. Definitionsof suicidal behavior: lessons learned from the WHo/EURO multicentre Study.Crisis 27, 4e15.

Diamond, G.S., Wintersteen, M.B., Brown, G.K., Diamond, G.M., Gallop, R., Shelef, K.,Levy, S., 2010. Attachment-based family therapy for adolescents with suicidalideation: a randomized controlled trial. J. Am. Acad. Child. Adolesc. Psychiatry49, 122e131.

Donaldson, D., Spirito, A., Esposito-Smythers, C., 2005. Treatment for adolescentsfollowing a suicide attempt: results of a pilot trial. J. Am. Acad. Child. Adolesc.Psychiatry 44, 113e120.

Dougherty, D.M., Mathias, C.W., Marsh-Richard, D.M., Prevette, K.N., Dawes, M.A.,Hatzis, E.S., Palmes, G., Nouvion, S.O., 2009. Impulsivity and clinical symptomsamong adolescents with non-suicidal self-injury with or without attemptedsuicide. Psychiatry Res. 169, 22e27.

Ducasse, D., Rene, E., Beziat, S., Guillaume, S., Courtet, P., Olie, E., 2014. Acceptanceand commitment therapy for management of suicidal patients: a pilot study.Psychother. Psychosom. 83, 374e376.

Duval, S., Tweedie, R., 2000. Trim and fill: a simple funnel-plot-based method oftesting and adjusting for publication bias in meta-analysis. Biometrics 56,455e463.

Egger, M., Smith, G.D., Phillips, A.N., 1997. Meta-analysis: Principles and procedures.BMJ 315, 1533e1537.

Erlangsen, A., Lind, B.D., Stuart, E.A., Qin, P., Stenager, E., Larsen, K.J., Wang, A.G.,Hvid, M., Nielsen, A.C., Pedersen, C.M., Winslov, J.H., Langhoff, C., Muhlmann, C.,Nordentoft, M., 2015. Short-term and long-term effects of psychosocial therapyfor people after deliberate self-harm: a register-based, nationwide multicentrestudy using propensity score matching. Lancet Psychiatry 2, 49e58.

Esposito-Smythers, C., Spirito, A., Kahler, C.W., Hunt, J., Monti, P., 2011. Treatment of

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e20 19

co-occurring substance abuse and suicidality among adolescents: a randomizedtrial. J. Consult Clin. Psychol. 79, 728e739.

Guthrie, E., Kapur, N., Mackway-Jones, K., Chew-Graham, C., Moorey, J., Mendel, E.,Marino-Francis, F., Sanderson, S., Turpin, C., Boddy, G., Tomenson, B., 2001.Randomised controlled trial of brief psychological intervention after deliberateself poisoning. Bmj 323, 135e138.

Hawton, K., Mckeown, S., Day, A., Martin, P., O'connor, M., Yule, J., 1987. Evaluationof out-patient counselling compared with general practitioner care followingoverdoses. Psychol. Med. 17, 751e761.

Hawton, K., Witt, K.G., Taylor Salisbury, T.L., Arensman, E., Gunnell, D., Hazell, P.,Townsend, E., Van Heeringen, K., 2015a. Pharmacological interventions for self-harm in adults. Cochrane Database Syst. Rev. 7, CD011777.

Hawton, K., Witt, K.G., Taylor Salisbury, T.L., Arensman, E., Gunnell, D., Townsend, E.,Van Heeringen, K., Hazell, P., 2015b. Interventions for self-harm in children andadolescents. Cochrane Database Syst. Rev. 12, CD012013.

Higgins, J.P.T., Green, S., 2011. Cochrane Handbook for Systematic Reviews of In-terventions Version 5.1.0 (updated March 2011). The Cochrane Collaboration.

Hollon, S.D., Derubeis, R.J., Evans, M.D., Wiemer, M.J., Garvey, M.J., Grove, W.M.,Tuason, V.B., 1992. Cognitive therapy and pharmacotherapy for depression.Singly and in combination. Arch. Gen. Psychiatry 49, 774e781.

Hvid, M., Vangborg, K., Sorensen, H.J., Nielsen, I.K., Stenborg, J.M., Wang, A.G., 2011.Preventing repetition of attempted suicideeII. The Amager project, a random-ized controlled trial. Nord. J. Psychiatry 65, 292e298.

Jadad, A.R., Moore, R.A., Carroll, D., Jenkinson, C., Reynolds, D.J., Gavaghan, D.J.,Mcquay, H.J., 1996. Assessing the quality of reports of randomized clinical trials:is blinding necessary? Control Clin. Trials 17, 1e12.

King, C.A., Klaus, N., Kramer, A., Venkataraman, S., Quinlan, P., Gillespie, B., 2009.The Youth-Nominated Support Team-Version II for suicidal adolescents: arandomized controlled intervention trial. J. Consult Clin. Psychol. 77, 880e893.

Klingberg, S., Herrlich, J., Wiedemann, G., Wolwer, W., Meisner, C., Engel, C., Jakobi-Malterre, U.E., Buchkremer, G., Wittorf, A., 2012. Adverse effects of cognitivebehavioral therapy and cognitive remediation in schizophrenia: results of thetreatment of negative symptoms study. J. Nerv. Ment. Dis. 200, 569e576.

Kuipers, E., Garety, P., Fowler, D., Dunn, G., Bebbington, P., Freeman, D., Hadley, C.,1997. London-East Anglia randomised controlled trial of cognitive-behaviouraltherapy for psychosis. I: effects of the treatment phase. Br. J. Psychiatry 171,319e327.

Linehan, M.M., 1997. Behavioral treatments of suicidal behaviors. Definitionalobfuscation and treatment outcomes. Ann. N. Y. Acad. Sci. 836, 302e328.

Linehan, M.M., Armstrong, H.E., Suarez, A., Allmon, D., Heard, H.L., 1991. Cognitive-behavioral treatment of chronically parasuicidal borderline patients. Arch. Gen.Psychiatry 48, 1060e1064.

Linehan, M.M., Comtois, K.A., Murray, A.M., Brown, M.Z., Gallop, R.J., Heard, H.L.,Korslund, K.E., Tutek, D.A., Reynolds, S.K., Lindenboim, N., 2006. Two-yearrandomized controlled trial and follow-up of dialectical behavior therapy vstherapy by experts for suicidal behaviors and borderline personality disorder.Arch. Gen. Psychiatry 63, 757e766.

Mcleavey, B.C., Daly, R.J., Ludgate, J.W., Murray, C.M., 1994. Interpersonal problem-solving skills training in the treatment of self-poisoning patients. Suicide LifeThreat Behav. 24, 382e394.

Mcmain, S.F., Links, P.S., Gnam, W.H., Guimond, T., Cardish, R.J., Korman, L.,Streiner, D.L., 2009. A randomized trial of dialectical behavior therapy versusgeneral psychiatric management for borderline personality disorder. Am. J.Psychiatry 166, 1365e1374.

Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., Group, P., 2009. Preferred reportingitems for systematic reviews and meta-analyses: the PRISMA statement. PLoSMed. 6, e1000097.

Morley, K.C., Sitharthan, G., Haber, P.S., Tucker, P., Sitharthan, T., 2014. The efficacy ofan opportunistic cognitive behavioral intervention package (OCB) on substanceuse and comorbid suicide risk: a multisite randomized controlled trial.J. Consult Clin. Psychol. 82, 130e140.

Muehlenkamp, J.J., Claes, L., Havertape, L., Plener, P.L., 2012. International preva-lence of adolescent non-suicidal self-injury and deliberate self-harm. Child.Adolesc. Psychiatry Ment. Health 6, 10.

Mufson, L., Dorta, K.P., Wickramaratne, P., Nomura, Y., Olfson, M., Weissman, M.M.,2004. A randomized effectiveness trial of interpersonal psychotherapy fordepressed adolescents. Arch. Gen. Psychiatry 61, 577e584.

Nordentoft, M., Jeppesen, P., Abel, M., Kassow, P., Petersen, L., Thorup, A., Krarup, G.,Hemmingsen, R., Jorgensen, P., 2002. OPUS study: suicidal behaviour, suicidalideation and hopelessness among patients with first-episode psychosis. One-year follow-up of a randomised controlled trial. Br. J. Psychiatry Suppl. 43,s98e106.

Ougrin, D., Boege, I., 2013. Brief report: the Self Harm Questionnaire: a new tooldesigned to improve identification of self harm in adolescents. J. Adolesc. 36,221e225.

Ougrin, D., Tranah, T., Leigh, E., Taylor, L., Asarnow, J.R., 2012. Practitioner review:

self-harm in adolescents. J. Child. Psychol. Psychiatry 53, 337e350.Ougrin, D., Tranah, T., Stahl, D., Moran, P., Asarnow, J.R., 2015. Therapeutic in-

terventions for suicide attempts and self-harm in adolescents: systematic re-view and meta-analysis. J. Am. Acad. Child. Adolesc. Psychiatry 54, 97 e2e107e2.

Perez, S., Marco, J.H., Garcia-Alandete, J., 2014. Comparison of clinical and de-mographic characteristics among borderline personality disorder patients withand without suicidal attempts and non-suicidal self-injury behaviors. Psychi-atry Res. 220, 935e940.

Pistorello, J., Fruzzetti, A.E., Maclane, C., Gallop, R., Iverson, K.M., 2012. Dialecticalbehavior therapy (DBT) applied to college students: a randomized clinical trial.J. Consult Clin. Psychol. 80, 982e994.

Rucci, P., Frank, E., Scocco, P., Calugi, S., Miniati, M., Fagiolini, A., Cassano, G.B., 2011.Treatment-emergent suicidal ideation during 4 months of acute managementof unipolar major depression with SSRI pharmacotherapy or interpersonalpsychotherapy in a randomized clinical trial. Depress Anxiety 28, 303e309.

Rudd, M.D., Bryan, C.J., Wertenberger, E.G., Peterson, A.L., Young-Mccaughan, S.,Mintz, J., Williams, S.R., Arne, K.A., Breitbach, J., Delano, K., Wilkinson, E.,Bruce, T.O., 2015. Brief cognitive-behavioral therapy effects on post-treatmentsuicide attempts in a military sample: results of a randomized clinical trial with2-Year follow-up. Am. J. Psychiatry 172 (5), 441e449 appiajp.201414070843.

Salkovskis, P.M., Atha, C., Storer, D., 1990. Cognitive-behavioural problem solving inthe treatment of patients who repeatedly attempt suicide. A controlled trial. Br.J. Psychiatry 157, 871e876.

Statsoft, I, 1995. STATISTICA for Windows [Online]. StatSoft Italia srl.Swannell, S.V., Martin, G.E., Page, A., Hasking, P., St John, N.J., 2014. Prevalence of

nonsuicidal self-injury in nonclinical samples: systematic review, meta-analysisand meta-regression. Suicide Life Threat Behav. 44, 273e303.

Sweeting, M.J., Sutton, A.J., Lambert, P.C., 2004. What to add to nothing? Use andavoidance of continuity corrections in meta-analysis of sparse data. Stat. Med.23, 1351e1375.

Tarrier, N., Taylor, K., Gooding, P., 2008. Cognitive-behavioral interventions toreduce suicide behavior: a systematic review and meta-analysis. Behav. Modif.32, 77e108.

Turner, B.J., Austin, S.B., Chapman, A.L., 2014. Treating nonsuicidal self-injury: asystematic review of psychological and pharmacological interventions. Can. J.Psychiatry 59, 576e585.

Tyrer, P., Thompson, S., Schmidt, U., Jones, V., Knapp, M., Davidson, K., Catalan, J.,Airlie, J., Baxter, S., Byford, S., Byrne, G., Cameron, S., Caplan, R., Cooper, S.,Ferguson, B., Freeman, C., Frost, S., Godley, J., Greenshields, J., Henderson, J.,Holden, N., Keech, P., Kim, L., Logan, K., Manley, C., Macleod, A., Murphy, R.,Patience, L., Ramsay, L., De Munroz, S., Scott, J., Seivewright, H., Sivakumar, K.,Tata, P., Thornton, S., Ukoumunne, O.C., Wessely, S., 2003. Randomizedcontrolled trial of brief cognitive behaviour therapy versus treatment as usualin recurrent deliberate self-harm: the POPMACT study. Psychol. Med. 33,969e976.

Van Der Sande, R., Van Rooijen, L., Buskens, E., Allart, E., Hawton, K., Van DerGraaf, Y., Van Engeland, H., 1997. Intensive in-patient and community inter-vention versus routine care after attempted suicide. A randomised controlledintervention study. Br. J. Psychiatry 171, 35e41.

Verheul, R., Van Den Bosch, L.M., Koeter, M.W., De Ridder, M.A., Stijnen, T., Van DenBrink, W., 2003. Dialectical behaviour therapy for women with borderlinepersonality disorder: 12-month, randomised clinical trial in The Netherlands.Br. J. Psychiatry 182, 135e140.

Victor, S.E., Klonsky, E.D., 2014. Correlates of suicide attempts among self-injurers: ameta-analysis. Clin. Psychol. Rev. 34, 282e297.

Vitiello, B., Silva, S.G., Rohde, P., Kratochvil, C.J., Kennard, B.D., Reinecke, M.A.,Mayes, T.L., Posner, K., May, D.E., March, J.S., 2009. Suicidal events in thetreatment for adolescents with depression study (TADS). J. Clin. Psychiatry 70,741e747.

Walkup, J.T., Albano, A.M., Piacentini, J., Birmaher, B., Compton, S.N., Sherrill, J.T.,Ginsburg, G.S., Rynn, M.A., Mccracken, J., Waslick, B., Iyengar, S., March, J.S.,Kendall, P.C., 2008. Cognitive behavioral therapy, sertraline, or a combination inchildhood anxiety. N. Engl. J. Med. 359, 2753e2766.

Wei, S., Liu, L., Bi, B., Li, H., Hou, J., Tan, S., Chen, X., Chen, W., Jia, X., Dong, G., Qin, X.,Liu, Y., 2013. An intervention and follow-up study following a suicide attempt inthe emergency departments of four general hospitals in Shenyang, China. Crisis34, 107e115.

Weinberg, I., Gunderson, J.G., Hennen, J., Cutter Jr., C.J., 2006. Manual assistedcognitive treatment for deliberate self-harm in borderline personality disorderpatients. J. Pers. Disord. 20, 482e492.

Wong, D.F., 2008. Cognitive behavioral treatment groups for people with chronicdepression in Hong Kong: a randomized wait-list control design. DepressAnxiety 25, 142e148.

World Health Organization, 2014. Preventing Suicide: a Global Imperative[Accessed 15 February 2016].

R. Calati, P. Courtet / Journal of Psychiatric Research 79 (2016) 8e2020