Critical appraisal of research

A Systematic Review and Meta-Analysis of Cognitive Behavioural Treatments for Suicidal and Self-Harm Behaviours in Adolescents

Réal Labelle, Louise Pouliot, and Alain Janelle Université du Québec à Montréal

Although cognitive– behavioural (CB) treatments are recognised as evidence-based interventions for depression and other disorders in adolescents, their efficacy in reducing suicidal and self-harm behav- iours in this group remains equivocal. First, a systematic review of the literature was carried out (N � 25 studies) on CB treatments for adolescents who presented suicidal ideation, had made suicide attempts, or engaged in self-harm. Results suggest that the scientific quality of past studies is suboptimal. Second, a meta-analysis of the pooled data of 14 CB treatment studies was conducted using a pre–post control group design. Results indicate a significant treatment effect in reducing suicidal ideation (n � 13/14 studies; g � �.40, 95% CI [�.30, .49], z � 7.95, p � .001). A significant effect was observed also with respect to self-harm (n � 8/14 studies; g � �.27, 95% CI [�.17, .38], z � 4.96, p � .001). However, no significant effect was found for suicide attempts (n � 6/14 studies; g � �.01, 95% CI [�.13, .14], z � .07, p � .94). The poor effect observed in this case could be due to low baseline prevalence of suicide attempts in most studies owing to active exclusion of adolescents at high risk for suicide.

Keywords: cognitive– behavioural, treatment, suicidal behaviours, self-harm, adolescent

Suicide is the second leading cause of youth death in Canada (Canadian Association for Suicide Prevention, 2014). In 2011, 198 Canadian adolescents died by suicide (Statistics Canada, 2014). Moreover, evidence dating back to 2008 (Kutcher & Szumilas, 2008) attributed 25% of youth hospital admissions to suicide attempts. In addition, a recent community survey revealed that 8% of the adolescents surveyed had seriously considered suicide in the previous 12 months (Fuller-Thomson, Hamelin, & Granger, 2013). Furthermore, it has been estimated that 18% of adolescents engage in nonsuicidal self-harm behaviours in their teens (Muehlenkamp, Claes, Havertape, & Plener, 2012). Although these behaviours are not driven by suicidal intent, they are nonetheless a recognised risk factor for adolescent suicidal behaviours (Cooper et al., 2005).

As defined by O’Connor and Nock (2014), suicidal behaviour is any self-reported thought of suicide and self-inflicted act with evidence of suicidal intent. Suicide ideation (SI) is the term used in practice to refer to any self-reported thought of engaging in suicide-related behaviour. Suicide attempt (SA) is the term widely used in North America for the full range of self-inflicted acts with the intent to kill oneself. Finally, the Canadian Patient Safety Institute and the Ontario Hospital Association (Perlman, Neufeld,

Martin, Goy, & Hirdes, 2011) defined self-harm (SH) as the deliberate and usually repetitive destruction or alteration of one’s own body tissue, without evidence of suicidal intent. Self-harm covers a wide range of behaviours including ingestion of toxic substances, burning, cutting, scratching, banging body parts, and hair pulling.

Despite the high prevalence of SI, SA, and SH among adoles- cents, there is yet no reliable information on how best to prevent the recurrence of these behaviours. On the grounds that mental illness is one of the major risk factors for suicidal and self-harm behaviours in adolescents (for reviews, see Berman, Jobes, & Silverman, 2005; Bursztein & Apter, 2011; Wagner, 2009), many psychosocial interventions have been developed and implemented to address mental health issues in adolescents with such behav- iours. From an empirical point of view, it was reasonable for clinical researchers to infer that the alleviation of symptoms re- lated to psychological disorders would suffice to prevent relapse of suicidal and self-harm behaviours among adolescents under care (Spirito, Esposito-Smythers, Wolff, & Uhl, 2011; Wagner, 2009).

In the past 15 years, a number of reviews have been published on the efficacy of psychosocial treatments in alleviating adolescent suicidal and self-harm behaviours (Brent et al., 2013; Brunstein Klomek & Stanley, 2007; Bursztein & Apter, 2011; Dubicka et al., 2010; Groves, Backer, van den Bosch, & Miller, 2012; Hawton, Saunders, & O’Connor, 2012; Klein & Miller, 2011; Ougrin, Tranah, Leigh, Taylor, & Asarnow, 2012; Pelkonen & Marttunen, 2003; Robinson, Hetrick, & Martin, 2011; Spirito et al., 2011; Tarrier, Taylor, & Gooding, 2008; Townsend et al., 2010; Wagner, 2009; Wasserman et al., 2012; Weisz, McCarty, & Valeri, 2006). Unfortunately, most of these (63%) were narrative reviews that summarised evidence qualitatively and, as such, did not employ a robust and replicable methodology. It is no surprise, then, that their conclusions have been contradictory at times. Some authors have found the evidence in support of the efficacy of psychosocial

This article was published Online First May 25, 2015. Réal Labelle, Louise Pouliot, and Alain Janelle, Département de psy-

chologie et Centre de recherche et d’intervention sur le suicide et l’euthanasie, Université du Québec à Montréal.

We thank Charles Cardinal, librarian at the Centre de recherche et d’intervention sur le suicide et l’euthanasie, and his associate Steve Car- rière, for their assistance in the literature searches.

Correspondence concerning this article should be addressed to Réal Labelle, Département de psychologie, Université du Québec à Montréal, Case postale 8888, succursale Centre-Ville, Montréal, Québec, H3C 3P8. E-mail: [email protected]

Canadian Psychology / Psychologie canadienne © 2015 Canadian Psychological Association 2015, Vol. 56, No. 4, 368 –378 0708-5591/15/$12.00 http://dx.doi.org/10.1037/a0039159

368

treatments in preventing the recurrence of suicidal and self-harm behaviours in adolescents to be equivocal (Brunstein Klomek & Stanley, 2007; Bursztein & Apter, 2011; Hawton et al., 2012; Pelkonen & Mattunen, 2003; Spirito et al., 2011; Wagner, 2009), whereas others have concluded that psychotherapies centred on support strategies (Brent et al., 2013; Ougrin et al., 2012) or on core emotion regulation skills (Groves et al., 2012; Klein & Miller, 2011) hold great promise. In addition, Wasserman and colleagues (2012), who also reviewed the literature on the effectiveness of psychosocial treatments, have contended that available empirical findings support the efficacy of cognitive– behavioural (CB) treat- ments in preventing the recurrence of suicidal and self-harm be- haviours in all clinical populations (i.e., including children and adolescents).

In addition to these narrative reviews, five quantitative reviews have also been published (Dubicka et al., 2010; Robinson et al., 2011; Tarrier et al., 2008; Townsend et al., 2010; Weisz et al., 2006). The Townsend et al. (2010) meta-analytical synthesis, which covered any type of psychosocial intervention, reported a nonsignificant improvement in self-harm based on the findings of a single trial (i.e., Rohde, Clarke, Mace, Jorgensen, & Seeley, 2004). Tarrier et al. (2008), for their part, pooled data on a mix of outcome variables (i.e., dissatisfaction with life, hopelessness, suicidal ideation, suicidal plans, suicide threats, and probability of suicide) from seven independent studies of controlled cognitive– behavioural therapy (CBT) trials; their findings revealed a nonsig- nificant effect size for the interventions (g � �.26, 95% CI [�.635, .116], z � �1.36, p � .18). The meta-analysis that Robinson et al. (2011) conducted on the basis of randomized controlled trials of an array of psychosocial interventions did not find that active treatments produced superior SI and SA outcomes compared with control conditions. Closer inspection of this study revealed, however, that the analysis was performed on the data of only two trials. In their meta-analysis, Dubicka and colleagues (2010) addressed the issue of whether addition of CBT to antide- pressants reduced suicidal behaviour in depressed adolescents more than did antidepressants alone. These authors considered four randomized controlled trials with different outcome variables (i.e., SI, SA, and SH). They reported no significant difference between the combined treatment (CBT and antidepressants) and antidepres- sants alone (g � �.05, 95% CI [�.18, .28], p � .66). In the fifth meta-analysis, Weisz et al. (2006) focused on six controlled stud- ies of depressed youth, all involving CBT and a measure of suicidal behaviour (e.g., total of suicide symptoms from a diag- nostic interview or a questionnaire designed to tap suicidal think- ing and behaviours in depressive youth). They reported a small and nonsignificant mean effect size for suicidal behaviours (g � �.18, z � 1.80, p � .07).

On the whole, these meta-analyses have provided very different estimates of the benefits of psychosocial treatments in terms of relapse prevention for suicidal and self-harm behaviours in ado- lescents. This could be explained in part by various factors, including insufficient statistical power (Dubicka et al., 2010; Robinson et al., 2011; Tarrier et al., 2008; Townsend et al., 2010; Weisz et al., 2006), varying and/or loose definitions of treatment outcomes (Tarrier et al., 2008), and different study inclusion/ exclusion criteria with, in some cases, a lack of specificity in the criteria chosen. Age range of participants is another point of interest that is worth addressing with respect to these meta-

analyses. Some of the studies included in the analyses had been conducted with participants as young as 11 years of age (e.g., Dubicka et al., 2010) and as old as 25 years (e.g., Robinson et al., 2011). From a developmental perspective, 11-year-olds (prepuber- tal children) and 25-year-olds (young adults) do not cognitively appraise and emotionally respond to their environment as adoles- cents do.

Finally, these early qualitative and quantitative literature re- views suggest that, at best, psychosocial treatments, including those applying a CB approach, have limited efficacy in reducing suicidal and self-harm behaviours in adolescents. Conclusions of previous scholars’ works nonetheless have to be put in the context of procedural, conceptual, and quantitative issues forwarded.

Objectives of This Project

Given the equivocal picture painted by existing narrative and meta-analytic synthesis of the effectiveness of CB treatments for suicidal and self-harm behaviours in adolescents and given the availability of new research in this area, we undertook to complete first a systematic review of the literature in this area, followed by meta-analysis of the results of independent studies to determine whether CB treatments are effective.

The aims of this project were fourfold. First, as research had progressed and new evidence had become available, we sought to update the earlier reviews. Second, we sought to evaluate and document in a systematic manner the potential methodological drawbacks of the existing literature, which might limit the infer- ences drawn about the therapeutic value of CB treatments in preventing the relapse of suicidal and self-harm behaviours. Third, we aimed to determine whether CB interventions have differential therapeutic values (mean estimated effects) by type of suicidal and self-harm behaviours (i.e., SI, SA, SH). Fourth and last, we in- tended to assess whether duration, number of therapeutic sessions, number of intervention modalities (i.e., family sessions vs. group sessions vs. individual sessions) employed during treatment, and overall quality of study were significant correlates of treatment effects on suicidal and self-harm behaviour outcomes across stud- ies.

The Systematic Review

Method

Search strategy. We used two techniques to identify CBT outcome studies. First, the Cochrane Central Register of Con- trolled Trials (CENTRAL), MEDLINE, PsycINFO, and Disserta- tion Abstracts electronic databases were searched between January 1, 1990, and March 31, 2014, for potentially relevant published research reports. The following terms were used to find papers reporting clinical trials: (cognitivo-behavioural OR cognitivo- behavioural OR “cognitive behavioural” OR cognitive– behavioural OR CBT) AND (adolescen� OR youth OR teen�) AND (depress� OR suicid� OR “self-harm” OR self-injur�) AND (treatment� OR therap� OR psychotherap� OR counsel�) AND (“case-control” OR trial OR experiment� OR “randomized con- trolled trial”). Second, the reference lists of relevant study reports and literature reviews were sifted through for articles of potential interest.

369EVIDENCE-BASED PRACTICE

Inclusion criteria. We established a clear set of criteria to determine what research on the effects of CB treatments on sui- cidal and self-harm behaviour outcomes in adolescents to include in the systematic review. First, selection was limited to studies conducted with participants ranging in age from 12 to 18 years and under clinical care at study entry. Second, only studies that tested CB treatment efficacy in preventing relapse of SI, SA, or SH were retained. Third, studies were included in the systematic review only if they used a pre–post design and reported original data. Outcome studies published in English, French, or German were considered. During the search process, only one research report published in German was located for which the assistance of a German translator was sought. However, on further investigation, we realised this article had been published again 5 years later, in English, in a well-known American scientific journal. We kept the English version of the article for our analyses. Finally, studies had to have been conducted using a pre–post control group design and had to have reported enough statistics to allow conversion to a standardized effect metric.

Exclusion criteria. Studies examining CB treatments to re- duce suicidal and self-harm behaviours in intellectually disabled adolescents, CB interventions dispensed in a school context, and CB treatments delivered online to adolescents were excluded, as were studies in which adolescents were not the main target of the intervention and those not reviewed by a committee of peers.

Selection of studies and data extraction. Two of the authors independently examined all potentially relevant studies for inclu-

sion in the review. Any disagreement between them was resolved through discussion with the third author. Inclusion assessment was performed using a standardized data extraction form and took into account information on type of article, type of study design, characteristics of participants, type of intervention, and outcome variables.

Assessment of study quality. The methodological quality of the studies included in the review was assessed using a 21-item scale that covered research aims, design, population, definition and ascertainment of outcomes, and statistical analysis. Table 1 lists the items included in our quantitative scale. The studies were rated separately by two of the authors and any disagreement was re- solved by consensus through discussion with the third author. Agreement between the researchers proved adequate as evidenced by a kappa of .70 at p � .05. A 27-point quality assessment scale with a score range of 3 to 27 was created. Item development drew on the Cochrane Collaboration Handbook criteria (Higgins & Green, 2011) and on the Tarrier and Wykes (2004) quality assess- ment instrument for the evaluation of psychological treatment studies.

Results

Sample of studies. The initial search yielded 1,657 articles. Of these, 341 were excluded for being published in non–peer- reviewed journals. Based on close examination of the titles and abstracts of the remaining 1,316 articles, we excluded 1,291 for

Table 1 Assessment Scale of Methodological Quality of Design and Execution of Studies

No. Item label Numerical value

1 Is the objective of the study clearly described? 0 � no, 1 � yes 2 Is the hypothesis/research question clearly defined? 0 � no, 1 � yes 3 What is the type of study design? 1 � pre–post single group, 2 � pre–post

control group, 3 � randomized control trial

4 Was the number of participants calculated in advance for statistical power? 0 � no, 1 � yes 5 Did the study have sufficient power to detect the clinical important effect at 1 �n� � .80,

� � .05? 0 � no, 1 � yes

6 Was the patients’ clinical profile in different intervention groups similar? 0 � no, 1 � yes 7 Were the enrolled patients representative (based on inclusion and exclusion criteria used) of

the usual clientele met in clinical practice? 0 � no, 1 � yes

8 Was the treatment condition assignment concealed from both patients and treatment providers?

0 � no, 1 � yes or N/A mention if score of 1 obtained under Item 3

9 Was outcomes assessment carried out blind to treatment conditions? 0 � no, 1 � yes 10 Have the characteristics of patients who drop out been described? 0 � no, 1 � yes 11 Describe the level of specificity and psychometric properties of the outcomes’

measurement. 1 � weak, 2 � satisfactory, 3 �

excellent 12 Did the analysis include all enrolled patients (intent-to-treat analysis)? 0 � no, 1 � yes 13 Describe the overall level of theoretical and empirical justification of the experimental

intervention. 1 � weak, 2 � satisfactory, 3 �

excellent 14 Was a treatment protocol or manual used in the delivery of the experimental treatment? 0 � no, 1 � yes 15 Was a 1-year follow-up performed after the end of the intervention? 0 � no, 1 � yes 16 Can the results be generalized outside the study context? 0 � no, 1 � yes 17 Were the expected results of the DVs specified in the report? 0 � no, 1 � yes 18 Were the DVs operationalized and defined in the report? 0 � no, 1 � yes 19 Are the results of the DV reported? 0 � no, 1 � yes 20 Was there adequate control in the fluctuation of DV measurement? 0 � no, 1 � yes 21 Was there adequate control of the therapists’ fidelity/adherence to treatment

protocol/manual? 0 � no, 1 � yes

Note. DV � dependent variable. Total score could vary from 3 to 27.

370 LABELLE, POULIOT, AND JANELLE

failing to meet the inclusion criteria. Twenty-five studies were found to be suitable for inclusion in the systematic review portion of our study. Figure 1 presents a flowchart of the inclusion and exclusion criteria used.

Study characteristics. Table 2 presents selected characteris- tics of all 25 studies included in our systematic review. From the outset, it was clear that the selected studies theoretically fell into two subgroups: those that examined CB treatments that placed a greater emphasis on cognitive and behavioural elements (discussed below as CBT) and those that examined CB treatments with a focus instead on emotional management (discussed below as dialectical– behavioural therapy or DBT).

For the studies as a whole, the experimental treatment involved 47 adolescent patients on average, (SD � 48.56, range: 5–179). Mean age of patients was 15.52 years (SD � 0.55). Samples were composed predominantly of adolescent girls (79%). Ten of the studies (40%) used comorbid suicidal and self-harm behaviours as an entry criterion for participants. In 14 of the studies (56%), the principal enrolment criterion for participants was presence of mood disorder. In six of the studies (24%), participants presenting

“active suicidal behaviour” were excluded. The majority of studies that used this exclusion criterion concerned CBT.

Location and treatment provider. In nine studies (36%), treatment was delivered in an outpatient clinic setting. Thirteen studies were conducted in the United States (52%) and eight were conducted in the United Kingdom (32%). The main treatment providers were mostly licensed psychotherapists (88%), such as psychologists and psychiatrists, followed by social workers (32%).

Quality of studies. No study obtained a perfect score of 27 on the quality assessment scale. Overall, scores were rather low (M � 13.48, SD � 3.37, range: 8 –20). On average, CBT studies (M � 14.56, SD � 3.13) scored higher than DBT studies (M � 10.71, SD � 2.29), and the difference was statistically significant, t(23) � 2.94, p � .007, 95% CI [�6.544, 1.138]. Six studies (24%) evaluated pre- and posttreatment suicidal and self-harm behaviour outcomes with no control group, and seven (28%) used treatment as usual (TAU) as the control condition. Single-subject designs were used mainly in DBT studies (see Table 2).

Overall, only two studies (8%) had sufficient statistical power to detect a significant effect of the experimental treatment on suicidal

Figure 1. Flowchart of inclusion and exclusion criteria. CB � cognitive– behavioural.

371EVIDENCE-BASED PRACTICE

T ab

le 2

S el

ec te

d C

h a ra

ct er

is ti

cs o f

S tu

d ie

s o f

U se

o f

C B

T a n d

D B

T W

it h

A d o le

sc en

ts

S tu

dy N

S et

ti ng

an d

pa rt

ic ip

an ts

C on

tr ol

D ur

at io

n/ nu

m be

r of

se ss

io ns

O ut

co m

e m

ea su

re s

d a

B re

nt et

al .

(1 99

7) †

10 7

O ut

pa ti

en t

cl in

ic ,

ag es

13 –1

8 ye

ar s,

76 %

F S

ys te

m ic

be ha

vi ou

r fa

m il

y th

er ap

y/ no

nd ir

ec ti

ve su

pp or

ti ve

tr ea

tm en

t

12 –1

6 W

K -S

A D

S s

� 0.

36 c

B re

nt et

al .

(2 00

8) †

33 4

U ni

ve rs

it y

an d

co m

m un

it y

cl in

ic ,

ag es

12 –1

8, ye

ar s,

70 %

F V

en la

fa xi

ne /S

S R

I 12

W &

12 S

S IQ

, N

S 0.

02 b ,0

.2 1c

,0 .0

9d

B re

nt et

al .

(2 00

9) †

12 4

S et

ti ng

N S

, ag

es 12

–1 8

ye ar

s, 72

% F

C B

T �

m ed

ic at

io n/

m ed

ic at

io n

26 W

S S

I, C

S H

F N

/A C

la rk

e et

al .

(2 00

2) †

88 S

et ti

ng N

S ,

m ea

n ag

e �

15 .3

ye ar

s, 68

% F

T A

U 8

W &

16 S

K -S

A D

S –P

L �

0. 05

b ,�

0. 05

c

D on

al ds

on ,

S pi

ri to

, &

E sp

os it

o- S

m yt

he rs

(2 00

5) †

31 O

ut pa

ti en

t cl

in ic

, ag

es 12

–1 7

ye ar

s, 82

% F

S up

po rt

iv e

re la

ti on

sh ip

tr ea

tm en

t 12

W &

7 S

S IQ

, C

I �

0. 29

b ,0

.5 2c

E sp

os it

o- S

m yt

he rs

, S

pi ri

to ,

U th

, &

L aC

ha nc

e (2

00 6)

† 6

O ut

pa ti

en t

cl in

ic ,

m ea

n ag

e �

15 ye

ar s,

83 %

F N

on e

26 W

& 31

S S

IQ ,

N S

� 1.

79 b ,�

0. 01

c

E sp

os it

o- S

m yt

he rs

, S

pi ri

to ,

K ah

le r,

H un

t &

M on

ti (2

01 1)

† 40

O ut

pa ti

en t

cl in

ic ,

ag es

13 –1

7 ye

ar s,

68 %

F T

A U

48 W

& 33

S K

-S A

D S

–P L

, S

IQ �

0. 04

b ,�

0. 82

c

F le

is ch

ha ke

r et

al .

(2 01

1) ‡

12 S

et ti

ng N

S ,

ag es

13 –1

9 ye

ar s,

10 0%

F N

on e

16 –2

4 W

& 32

S L

P C

0. 00

c ,�

0. 43

d

G oo

dy er

et al

. (2

00 7)

† 20

8 O

ut pa

ti en

t cl

in ic

, ag

es 11

–1 7

ye ar

s, %

F N

S T

A U

28 W

& 28

S K

-S A

D S

–P L

� 0.

17 b ,�

0. 17

d

G re

en et

al .

(2 01

1) †

36 6

S et

ti ng

N S

, ag

es 12

–1 7

ye ar

s, 89

% F

T A

U N

S C

I, S

IQ �

1. 31

b ,�

0. 52

c

H ar

ri ng

to n

et al

. (1

99 8)

† 16

2 S

et ti

ng N

S ,

ag es

10 –1

6 ye

ar s,

87 %

F T

A U

5 W

& 5

S S

IQ ,

D S

H IS

� 0.

13 b ,0

.0 0d

H az

el l

et al

. (2

00 9)

† 72

S et

ti ng

N S

, ag

es 12

–1 6

ye ar

s, 91

% F

T A

U 52

W &

6 S

S IQ

, C

I �

0. 09

b ,�

0. 25

d

Ja m

es ,

T ay

lo r,

W in

m il

l, &

A lf

oa da

ri (2

00 8)

‡ 16

S et

ti ng

N S

, ag

es 15

–1 8

ye ar

s, 10

0% F

N on

e 52

W &

10 4

S C

I �

1. 42

d

Ja m

es ,

W in

m il

l, A

nd er

so n,

& A

lf oa

da ri

(2 01

1) ‡

25 S

et ti

ng N

S ,

ag es

13 –1

7 ye

ar s,

88 %

F N

on e

52 W

& 10

4 S

C I

� 0.

62 d

K at

z, C

ox ,

G un

as ek

ar a,

& M

il le

r (2

00 4)

‡ 62

In pa

ti en

t cl

in ic

, ag

es 14

–1 7

ye ar

s, 84

% F

P sy

ch od

yn am

ic tr

ea tm

en t

2 W

& 10

S S

IQ ,

L P

C �

0. 08

b ,�

0. 20

c ,d

M ar

ch et

al .

(2 00

4) †

43 9

C om

m un

it y

cl in

ic ,

ag es

12 –1

7 ye

ar s,

53 %

F T

A U

/p la

ce bo

12 W

& 12

S S

IQ ,

C I

� 0.

34 b ,0

.1 1c

, �

0. 13

d

M cD

on el

l et

al .

(2 01

0) ‡

21 0

In pa

ti en

t cl

in ic

, ag

es 12

–1 7

ye ar

s, 58

% F

R ec

re at

io na

l in

te rv

en ti

on N

S M

R A

� 0.

41 d

M el

vi n

et al

. (2

00 6)

† 73

C om

m un

it y

cl in

ic ,

ag es

12 –1

8, 65

% F

C B

T �

m ed

ic at

io n/

m ed

ic at

io n

12 W

& 12

S S

IQ �

0. 36

b

O ld

er sh

aw et

al .

(2 01

2) †

55 C

om m

un it

y cl

in ic

, ag

es 12

–1 8

ye ar

s, 96

% F

N o

tr ea

tm en

t/ he

al th

y su

bj ec

ts 12

W &

15 S

H M

Q �

1. 31

d

R at

hu s

& M

il le

r (2

00 2)

‡ 11

1 O

ut pa

ti en

t cl

in ic

, m

ea n

ag e

� 16

.1 ye

ar s,

93 %

F P

sy ch

od yn

am ic

/s up

po rt

iv e

th er

ap y

12 W

& 24

S S

S I,

C I

� 1.

20 b ,0

.0 0c

R oh

de et

al .

(2 00

4) †

93 S

et ti

ng N

S ,

ag es

13 –1

7 ye

ar s,

48 %

F L

if e

sk il

ls tu

to ri

ng 8

W &

16 S

C I

N /A

T ay

lo r

et al

. 20

11 )†

25 C

om m

un it

y cl

in ic

, ag

es 12

–1 8

ye ar

s, %

F N

S N

on e

W N

S &

8– 12

S S

H I–

R �

1. 55

d

V it

ie ll

o et

al .

(2 00

9) †

12 4

S et

ti ng

N S

, ag

es 12

–1 8

ye ar

s, 96

% F

C B

T �

m ed

ic at

io n/

m ed

ic at

io n

26 W

& 22

S S

S I

N /A

W oo

d, T

ra in

or ,

R ot

hw el

l, M

oo re

, &

H ar

ri ng

to n

(2 00

1) †

63 S

et ti

ng N

S ,

ag es

12 –1

8 ye

ar s,

78 %

F T

A U

26 W

& 6

S S

IQ ,

C I

0. 16

b ,�

1. 10

d

W oo

db er

ry &

P op

en oe

(2 00

8) ‡

28 S

et ti

ng N

S ,

ag es

12 –1

8 ye

ar s,

78 %

F N

on e

16 W

& 6–

28 S

T S

C C

, C

B C

L �

0. 73

b ,�

0. 62

c ,d

N o te

. †

C B

T �

co gn

it iv

e– be

ha vi

ou ra

l th

er ap

y; ‡

D B

T �

di al

ec ti

ca l–

be ha

vi ou

ra l

th er

ap y;

F �

fe m

al e;

N S

� no

t sp

ec if

ie d;

N /A

� no

t ap

pl ic

ab le

; C

B C

L �

C hi

ld B

eh av

io r

C he

ck li

st (A

ch en

ba ch

, 19

91 );

C D

R S

–R �

C hi

ld ho

od D

ep re

ss io

n R

at in

g S

ca le

— R

ev is

ed (P

oz na

ns ki

& M

ok ro

s, 19

95 );

C I

� cl

in ic

al in

te rv

ie w

; C

S H

F �

C ol

um bi

a S

ui ci

de H

is to

ry F

or m

(M an

n et

al .,

19 92

); D

S H

IS �

D el

ib er

at e

S el

f- H

ar m

In te

rv ie

w S

ch ed

ul e

(K er

fo ot

, 19

88 );

H M

Q �

H om

e- M

ad e

Q ue

st io

nn ai

re ;

K -S

A D

S –P

L �

S ch

ed ul

e fo

r A

ff ec

ti ve

D is

or de

rs an

d S

ch iz

op hr

en ia

fo r

S ch

oo l-

A ge

C hi

ld re

n— P

re se

nt an

d L

if et

im e

V er

si on

(K au

fm an

et al

., 19

97 );

L P

C �

L if

et im

e P

ar as

ui ci

de C

ou nt

(L in

eh an

& C

om to

is ,1

99 4)

; M

R A

� M

ed ic

al R

ec or

d A

na ly

si s;

S �

se ss

io ns

; S

H I–

R �

S el

f- H

ar m

In ve

nt or

y— R

ev is

ed fo

r A

do le

sc en

ts (D

av id

so n

et al

., 20

06 );

S IQ

� S

ui ci

da l

Id ea

ti on

Q ue

st io

nn ai

re (R

ey no

ld s,

19 88

; R

ey no

ld s

& M

az za

, 19

99 );

S S

I �

S ca

le fo

r S

ui ci

de Id

ea ti

on (B

ec k,

K ov

ac s,

& W

ei ss

m an

, 19

79 );

T S

C C

� T

ra um

a S

ym pt

om s

C he

ck li

st fo

r C

hi ld

re n

(B ri

èr e,

19 96

); S

S R

I �

se le

ct iv

e se

ro to

ni n

re up

ta ke

in hi

bi to

rs ;

T A

U �

tr ea

tm en

t as

us ua

l; W

� w

ee ks

. a

S ta

nd ar

di se

d ef

fe ct

si ze

. b

S ui

ci da

l id

ea ti

on .

c S

ui ci

de at

te m

pt .

d S

el f-

ha rm

.

372 LABELLE, POULIOT, AND JANELLE

and self-harm behaviours. We estimated that studies had to be conducted with at least 144 participants per condition to be able to detect an effect size set beforehand for our purposes at d � �0.33, with a power of 1 � � � .20 and � � .05. The effect size used to compute this estimate was taken from Tarrier and colleagues’ (2008) meta-analytic result.

There was substantial variability across studies in how key treatment outcomes were assessed (for details, see Table 2). De- scriptive data indicated that 13 of the 25 studies (50%) employed at least one measure with limited reliability, validity, specificity, and sensitivity information available to support its use in assessing change in suicidal or self-harm behaviours. Furthermore, four studies (11%) provided no information on how treatment outcomes were rated.

Of the 25 studies included in our review, six (24%) assessed outcomes at 1-year follow-up and 21 were conducted using manual-based interventions (84%). In 12 (48%), assessors might not have been blinded to the treatment condition to which patients were allocated. Fifteen studies (60%) indicated that all patients enrolled at inception were included in the analyses (i.e., intent-to- treat analyses). Eleven (40%) clearly stated that treatment provided by therapists to adolescents was controlled (e.g., videotaped ses- sions) to ensure greater standardization of treatment delivery across participants. Six studies (24%) mentioned that an adequate control procedure (e.g., training to study research staff on assess- ment tools, assessment repeated by a second member of research team) was implemented to prevent potential biases in the evalua- tion of study outcomes. Further observations revealed that 16 empirical reports (64%) presented study objectives clearly enough and that 10 (40%) were specific about the hypotheses and/or research questions addressed.

The Meta-Analysis

Our meta-analytic review focused on estimating the mean effect size of CB treatments on suicidal and self-harm behaviours and its statistical significance.

Method

Sample of studies. Of the 25 studies that were included in the systematic review, only 14 studies met all the criteria for inclusion in the meta-analysis. Single-group designs (n � 11) were excluded as these tend to generate inflated effect sizes that are not compa- rable to those yielded by control group design studies. The stan- dardized effect size values for the individual studies are given in Table 2.

Effect size calculation and data analysis procedures. Cohen’s d (1988) was used to measure effect size in the raw data analyses. As a preliminary data inspection revealed substantial variability in sample size across studies, unbiased effect sizes were calculated also through the procedure described by Hedges and Olkin (1985), whereby each effect size is weighted by sample size. Effect sizes were calculated so that a negative score indicated that the experimental treatment group improved more than the control group. A positive score indicated the opposite outcome. The raw effect estimates for all studies were pooled to obtain a raw com- bined effect size. The same was done for weighted effect size estimates. Furthermore, at the data extraction stage, it appeared

that the effect sizes obtained might have been influenced by various sources of considerable variability, including number and type of treatment modalities, number of sessions and/or length of treatment, and source of outcome measures. Consequently, for all our meta-analyses, we used the random effects model with a 95% confidence interval—a conservative statistical approach that takes the heterogeneousness of studies into account. Publication bias in favour of significant results was assessed using the statistical procedure described by Rosenthal and Rosnow (1991). Relation- ships between study effect size and rating, on the one hand, and treatment characteristics, on the other, were investigated using Pearson’s r (two-tailed tests).

Analyses were run using the statistical functions in Excel 2010 and mathematical formula macros designed by the authors with Excel (see: Rosenthal & Rosnow, 1991; Rosenthal & Rubin, 1982a, 1982b; Wolf, 1986). To maximize statistical power, we conducted the meta-analyses on data reported at posttreatment. Too few studies performed a follow-up assessment on outcome measures of interest at 1-year posttreatment.

Results

We pooled data by outcome of interest. Accordingly, the data from 13 studies were pooled for SI; there were six for SA and eight for SH. The number of studies varies based on the inclusion, in each study, of the outcome of interest. As all of these studies had used a pre–post control group design, their findings could be aggregated for the purposes of analysis.

Treatment effects on SI. The weighted mean effect (pooled estimate) of CBT/DBT on SI, based on 13 independent samples and 1,605 adolescent patients, was of moderate size (g � �.40, 95% CI [�.30, .49]) and statistically significant (z � 7.95, p � .001). In other words, this type of treatment had a positive and significant impact on SI. The Qt statistic calculated indicated that effect sizes were heterogeneous across the pooled findings of the 13 studies, Qt(12) � 99.51, p � .001.

Treatment effects on SA. The six studies that examined the effects of CBT/DBT on SA included 800 adolescents. The weighted mean effect size did not prove significant (g � �.01, 95% CI [�.13, .14], z � 0.07, p � .94). The homogeneity test revealed that effect sizes were heterogeneous, Qt(5) � 16.17, p � .01.

Treatment effects on SH. The eight studies pooled for SH included 1,305 adolescents. The weighted mean effect of treatment in this regard proved significant (g � �.27, 95% CI [�.17, .38], z � 4.96, p � .001). The null hypothesis of consistency in the results of these studies was rejected, Qt(7) � 29.57, p � .001.

Correlations between effect sizes, treatment, and study characteristics. Results showed a correlation between number of therapeutic sessions and treatment effect size on SI, r(15) � .54, p � .04. However, this treatment characteristic was not associated with effect sizes on SA or SH. Treatment duration, in terms of total weeks, was significantly associated with effect size on SA, r(7) � .83, p � .02. This was not the case for SI or SH. The bivariate correlation between quality of study and SH effect size proved negative and significant, r(15) � �.57, p � .03. Effect sizes on SI and SA were not significantly associated with study quality. In our correlation analyses, we further examined whether number of

373EVIDENCE-BASED PRACTICE

treatment modalities was related to change in SI, SA, or SH. No significant correlation emerged in this regard.

Publication bias. Based on Orwin’s (1983) equation, we es- timated the number of additional studies needed in the meta- analysis to reduce the averaged effect sizes obtained to d � 0.20 (our criterion value) at a p value of .01. Fail-safe N analyses revealed that for SI, it would have taken about 13 more studies and for SH, about three more with a d � 0.20, to dilute the significant meta-analytic effects reported. For SI, it was unlikely that a pub- lication bias distorted reported findings in any substantial way.

Discussion

Overall, this systematic review of the literature revealed that the evidence obtained in studies of the efficacy of CB treatments presented a number of limitations. Furthermore, our meta-analysis indicates that CB treatments are effective in reducing suicidal ideation and self-harm behaviours. Findings were mixed regarding associations between treatment characteristics, quality of study design, and effects on suicidal and self-harm behaviour outcomes.

Study Methodological Characteristics

We uncovered a number of methodological issues with the literature that was available. Although each study contained well- designed and methodologically sound elements, none obtained a perfect score of 27 on our quality assessment scale. Furthermore, many studies were unclear at the outset about their main objectives and/or the expected outcomes of the intervention tested. Without explicit statements concerning both objectives and hypotheses, it is hard to reach clear conclusions about the scientific and clinical implications of the reported findings. Our analysis suggested also that some studies were not sufficiently informed by theory and early empirical evidence. Theoretical models and previous empir- ical findings provide essential information about the causal mech- anisms behind effects.

We also found problems with the quality of the outcome mea- sures employed in the studies we reviewed. In many cases, the robustness of study results was potentially compromised by the use of outcome measures for which limited psychometric evidence of reliability and criterion-related validity was available. Conse- quently, some findings could owe more to the lack of validity and specificity of the measures used to assess the outcomes than to the actual effects of the interventions in question. It should be men- tioned, also, that for some of the studies reviewed, suicidal and self-harm behaviours were not the primary target of the interven- tion that was being assessed. Issues surrounding outcome mea- surements included cases in which the assessors were not blinded to the intervention condition, which clearly jeopardized the valid- ity of results.

In addition, for a number of studies, research design and imple- mentation did not meet the scientific standards to allow addressing possible threats to causal inference or proposing alternative expla- nations for the null, reverse, or positive effects reported. For example, the assessment of adherence to the intervention protocol was more an exception than the rule. Moreover, group equivalence in terms of clinical status was not demonstrated in most studies. Indeed, there were strong indications to the effect that patients were not sufficiently matched in terms of severity of psychopatho-

logical symptoms and comorbid mental health problems at base- line. For the majority of the studies under review, adequate statis- tical power was not attained to allow testing CBT/DBT efficacy in reducing suicidal behaviours, as these events are rare even in clinical samples.

Furthermore, although a good number of studies used an exper- imental randomized control group design, their results were none- theless vulnerable to numerous potential confounding factors. For example, the typical control condition against which CB treatment effectiveness was measured was TAU rather than another well- documented form of psychotherapy. This makes it difficult to determine, for example, whether the effect of the CB treatment was blurred by therapeutic elements of TAU similar to the exper- imental condition.

Lastly, none of the studies controlled for negative psychosocial stressors that might have arisen in the lives of the adolescent patients during the investigation. Negative life events that occur over the course of a study can reduce the degree of clinical improvement associated with treatment. Another issue concerned lack of control over professional profile (i.e., trainee psychologist or psychiatrist, social worker, nurse, licensed clinician) and level of expertise of those delivering the treatment. The considerable differences in findings could potentially be explained by this factor. Finally, in our quantitative review, we were unable to estimate the long-term effects of CBT and DBT on suicidal and self-harm behaviours in adolescents owing to the fact that too few trials included any follow-up beyond the end of treatment. The clinical meaningfulness of an intervention depends on this out- come.

Effects on Suicidal and Self-Harm Behaviours

In addition to identifying the methodological limitations of the studies available, this study aimed to estimate the magnitude of CB treatments effects on suicidal and self-harm behaviours in adoles- cents. Unlike earlier meta-analyses, this one showed CB treat- ments to be associated with significant treatment effects on SI and SH in adolescents. The effect on SI reported in the studies con- sidered was significant and moderate in size (g � �.40). In comparison, although previous meta-analyses evidenced a positive effect of CB treatment, as we did, the mean effects reported ranged from negligible to small (g � �.05, g � �.18, and g � �.26) and were not statistically significant (Dubicka et al., 2010; Tarrier et al., 2008; Weisz et al., 2006). This difference in results could be accounted for by lack of power to detect a significant effect in previous works, less than optimal specificity of treatment out- comes tested, and variability in age range of adolescents included in study.

It needs to be underscored that most of the studies whose findings were pooled for our meta-analysis assessed SI outcomes with psychometrically sound (i.e., reliable and valid) instruments (e.g., Scale for Suicide Ideation: Beck, Kovaks, & Weissman, 1979; Suicidal Ideation Questionnaire: Reynolds, 1988). The in- struments employed in these studies were sensitive enough to accurately detect change in SI. There is another reason to be optimistic about the potential benefits of CB treatment with respect to SI in adolescents: Calculation of the fail-safe N showed that it would have taken almost 13 more studies confirming the null

374 LABELLE, POULIOT, AND JANELLE

hypothesis—the same number retrieved for this meta-analysis—to reverse the conclusion of a significant effect.

However, it did not emerge from our analysis that CB treatments have a significant impact on SA. The effect estimated in our meta-analysis proved negligible (g � �.01). It needs to be men- tioned, however, that at least three key factors might mitigate this finding. First, some outcome studies included depressed adoles- cents who initially responded poorly to intensive treatment regi- mens and who continued whatever type of treatment they were undergoing while participating in the study. In these cases, a potential “ceiling effect” was present. The fact that all the adoles- cents enrolled in these studies presented a high baseline suicide risk and that they had already received acute treatment moderated how much of a significant and positive effect the adjunct of CBT or DBT could have on SA. What is more, and perhaps most important, these studies might be marred by treatment/intervention contamination. Second, another group of outcome studies might have been subject to the opposite confounding factor, that is, a “floor effect.” These trials excluded adolescents “actively suicidal” and included participants with near-bottom SA levels at baseline. Third, how SA outcomes were assessed was less than optimal owing to the dubious or untested psychometric properties of the measures used. This could have distorted the effects reported. This subgroup of studies, however, primarily investigated the effects of psychotherapeutic treatments on depressive symptoms and global functioning in adolescents; examination of their effects on SA was an ancillary consideration.

The posttreatment mean effect of CB treatments on SH was significant (g � �.27), although not quite as robust when esti- mated with the fail-safe N (i.e., about three more studies with an effect of d � �0.20 would be sufficient to render the finding nonsignificant). At least three major factors might support the positive impact of CB treatments on SH in adolescents. First, the vast majority of outcome studies that reported a significant effect in this regard had SH as the principal criterion for adolescents to be included in the trial. Second, in most cases, the treatment under evaluation focused on reducing this behaviour. Finally, it is worth noting that the psychotherapy provided to adolescents was sup- ported by a treatment manual.

Association Between Treatment Characteristics, Study Design, and Outcomes

Regarding how suicidal and self-harm behaviour outcomes, study quality, and treatment characteristics might be related, we observed no association between overall study quality and SI or SA outcomes but a negative relationship between study quality and decreased SH. In other words, the poorer the study design, the greater the effect reported. This finding is not surprising for the simple reason that we included all studies in our correlation analysis regardless of whether their design included a comparison group. Many of the studies reviewed in fact had a pre–post single-group design. In this regard, it is well documented that clinically significant changes can occur in patients over time whether or not they actually undergo therapy, and factors such as maturation, passage of time, and placebo effect may account for these changes.

Contrary to received wisdom in the clinical field, there was no evidence to the effect that the greater the number of treatment

components, the better the outcomes. No association emerged between any of the suicidal and self-harm behaviour outcomes and this treatment characteristic. However, intensity of treatment, de- fined here as number of sessions and total duration in weeks, did appear to some extent to affect SI and SA outcomes. Clinically speaking, this makes sense. SI and SA grow out of difficulty coping with problems, poor social skills, maladaptive beliefs, cognitive distortions, negative expectations, and difficulty regulat- ing emotions. Most of these behaviours and cognitions are ac- quired or learned over a long period of time. It would be overly optimistic to expect a meaningful improvement in SA without spending a minimum amount of time addressing underlying mal- adaptive cognitions and developing new, long-standing behav- ioural competencies.

Scientific Implications

Given that existing efficacy trials of CBT and DBT, believed to be plausible treatments for youth suicidal and self-harm behav- iours, were not conducted according to optimal methodological standards, it is important that future studies apply the appropriate methodological procedures to ensure both the external and the internal validity of their findings. An excellent way to do this would be to conduct randomized controlled trials with sufficient statistical power and adequate control of potential confounding factors.

Many studies have reported potential benefits in using CBT or DBT to alleviate youth suicidal and self-harm behaviours, but it remains to be determined which treatment modality (i.e., group, individual and/or family) is most effective. Furthermore, where positive effects have been observed, it is unclear whether cognitive restructuring, problem-solving skills, mood-regulation strategy components, a combination of these, or some other nonspecified factor (e.g., attention from therapist, peer support in group therapy) is responsible for the effects. Dismantling studies would be a promising avenue to answer these questions. Another issue of great clinical and theoretical interest concerns the relative efficacy of CBT and DBT in decreasing SI, SA, and SH compared with a control intervention condition. Because of the variability noted across studies in terms of care setting, population, treatment struc- ture and format, study design, control conditions, outcome mea- sures, and therapist background, there is a pressing need for independent replication studies. Findings from such studies would surely add to the body of knowledge by providing more adequate efficacy evaluations and strengthening the empirical foundations of the psychosocial treatments.

Clinical Implications

Until recently, many treatments for adolescents with suicidal and self-harm behaviours have rested on the non– evidence-based assumption that interventions focused on alleviating depression or Axis I mental disorders would, once primary symptoms are in remission, decrease relapse risk for such behaviours. As our meta- analysis has shown, there is good reason to reconsider this assump- tion. Indeed, clinically significant improvements in suicidal and self-harm behaviours were more likely with interventions designed specifically to alter suicidal and self-harm behaviours.

In this regard, we suggest, as did Spirito et al. (2011), that treatment should target suicidal and self-harm behaviours directly

375EVIDENCE-BASED PRACTICE

for maximum benefit. Like Spirito and colleagues, we, too, believe that building coping skills and boosting affect regulation in ado- lescents are promising avenues for treatment. The development of effective coping and emotion-regulation skills should, as studies have already shown, not only result in a significant decrease in suicidal and self-harm behaviours among adolescents in treatment but also provide the means to choose constructive alternatives for action in the face of adversity. What is more, these skills would be useful throughout adolescence and into adulthood.

The idea that suicidal and self-harm behaviours should be the direct targets of treatment is slowly gaining traction in the scien- tific and clinical community. In the recently published Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013), suicidal behaviour and nonsuicidal self-injury are considered potential specific clinical syndromes. The DSM-5 also outlines a series of proposed criteria for these conditions on which further research is encouraged.

In the Western world, adolescent males outnumber their female peers in suicide statistics. This may be perplexing given the fact that 79% of the participants enrolled in the trials under review were female. What this tells us is that, on the one hand, we cannot assume CB treatments to be efficacious in reducing suicidal and self-harm behaviours in adolescent males and, on the other, much still needs to be done to find ways to get adolescent males to seek treatment. Future research needs to consider these issues very closely.

Conclusion

This is the first systematic review and meta-analysis of the efficacy of CB treatments in reducing suicidal and self-harm behaviours in adolescents to estimate mean effect sizes distinctly for SI, SA, and SH. It shows CB treatments to be associated with clinically significant improvements in SI and SH in adolescents presenting with these clinical problems. Further controlled trials on the effects of CB treatment on suicidal and self-harm behaviours are clearly needed given the methodological shortcomings ob- served in the published studies that we reviewed.

Finally, it is important to remember that clinical trials to deter- mine the efficacy of CB treatments in reducing youth suicidal and self-harm behaviours are still at an early stage of development. More studies are clearly needed and these need to be methodolog- ically more rigorous.

Résumé

Les traitements cognitivo-comportementaux (CB) sont parmi les interventions basées sur des connaissances scientifiques reconnues pour traiter la dépression et d’autres troubles chez les adolescents. Toutefois, leur efficacité à réduire les comportements suicidaires et autodestructeurs parmi cette population demeure équivoque. Les auteurs ont d’abord effectué une revue systématique de la littéra- ture (N � 25 études) sur les traitements CB pour adolescents présentant des idées suicidaires, ayant fait des tentatives de suicide ou ayant des comportements autodestructeurs. Les résultats sug- gèrent que la qualité scientifique des études antérieures était sous- optimale. Ensuite, a été réalisée une méta-analyse des données recueillies dans 14 études sur les traitements CB au moyen d’un groupe de contrôle pré et post. Les résultats indiquent un effet

significatif des traitements pour la réduction des idées suicidaires (n � 13/14 études; g � �0,40, 95 % CI [�0,30, 0,49], z � 7,95, p � 0,001). Un effet significatif a aussi été constaté en ce qui a trait aux comportements autodestructeurs (n � 8/14 études; g � �0,27, 95 % CI [�0,17, 0,38], z � 4,96, p � 0,001). Toutefois, aucun effet significatif n’a été constaté sur le plan des tentatives de suicide (n � 6/14 études; g � �0.01, 95 % CI [�0,13, 0,14], z � 0,07, p � 0,94). Cet effet modéré pourrait être attribuable à la faible prévalence de tentatives de suicide faisant partie des études, en raison de l’exclusion voulue d’adolescents à haut risque de suicide.

Mots-clés : cognitivo-comportemental, traitement, comportement suicidaire, autodestructeur, adolescent.

References

References marked with an asterisk indicate studies included in the meta-analysis.

Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4 –18 and 1991 profile. Burlington, VT: University of Vermont.

American Psychiatric Association. (2013). Diagnostic and statistical man- ual of mental disorders (5th ed.). Washington, DC: Author.

Beck, A. T., Kovacs, M., & Weissman, A. (1979). Assessment of suicidal intention: The Scale for Suicide Ideation. Journal of Consulting and Clinical Psychology, 47, 343–352. http://dx.doi.org/10.1037/0022-006X .47.2.343

Berman, A. L., Jobes, D. A., & Silverman, M. M. (2005). Adolescent suicide: Assessment and intervention (2nd ed.). Washington, DC: Amer- ican Psychological Association.

�Brent, D., Emslie, G., Clarke, G., Wagner, K. D., Asarnow, J. R., Keller, M., . . . Zelazny, J. (2008). Switching to another SSRI or to venlafaxine with or without cognitive behavioral therapy for adolescents with SSRI- resistant depression: The TORDIA randomized controlled trial. Journal of the American Medical Association, 299, 901–913. http://dx.doi.org/ 10.1001/jama.299.8.901

�Brent, D. A., Greenhill, L. L., Compton, S., Emslie, G., Wells, K., Walkup, J. T., . . . Turner, J. B. (2009). The Treatment of Adolescent Suicide Attempters study (TASA): Predictors of suicidal events in an open treatment trial. Journal of the American Academy of Child & Adolescent Psychiatry, 48, 987–996. http://dx.doi.org/10.1097/CHI .0b013e3181b5dbe4

�Brent, D. A., Holder, D., Kolko, D., Birmaher, B., Baugher, M., Roth, C., . . . Johnson, B. A. (1997). A clinical psychotherapy trial for adolescent depression comparing cognitive, family, and supportive therapy. Ar- chives of General Psychiatry, 54, 877– 885. http://dx.doi.org/10.1001/ archpsyc.1997.01830210125017

Brent, D. A., McMakin, D. L., Kennard, B. D., Goldstein, T. R., Mayes, T. L., & Douaihy, A. B. (2013). Protecting adolescents from self-harm: A critical review of intervention studies. Journal of the American Acad- emy of Child & Adolescent Psychiatry, 52, 1260 –1271. http://dx.doi.org/ 10.1016/j.jaac.2013.09.009

Brière, J. (1996). Professional manual for the Trauma Symptom Checklist for Children (TSCC). Odessa, FL: Psychological Assessment Resources.

Brunstein Klomek, A., & Stanley, B. (2007). Psychosocial treatment of depression and suicidality in adolescents. CNS Spectrums, 12, 135–144.

Bursztein, C., & Apter, A. (2011). Evidence-based prevention and treat- ment of suicidality in children and adolescents. In R. C. O’Connor, S. Platt, & J. Gordon (Eds.), International handbook of suicide prevention: Research, policy and practice (pp. 291–529). Chichester, UK: Wiley- Blackwell. http://dx.doi.org/10.1002/9781119998556.ch17

376 LABELLE, POULIOT, AND JANELLE

Canadian Association for Suicide Prevention. (2014). Youth and suicide. Retrieved from http://suicideprevention.ca/understanding/suicide-and- high-risk-groups/

�Clarke, G. N., Hornbrook, M., Lynch, F., Polen, M., Gale, J., O’Connor, E., . . . Debar, L. (2002). Group cognitive– behavioral treatment for depressed adolescent offspring of depressed parents in a health mainte- nance organization. Journal of the American Academy of Child & Adolescent Psychiatry, 41, 305–313. http://dx.doi.org/10.1097/ 00004583-200203000-00010

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Cooper, J., Kapur, N., Webb, R., Lawlor, M., Guthrie, E., Mackway-Jones, K., & Appleby, L. (2005). Suicide after deliberate self-harm: A 4-year cohort study. The American Journal of Psychiatry, 162, 297–303. http:// dx.doi.org/10.1176/appi.ajp.162.2.297

Davidson, K., Norrie, J., Tyrer, P., Gumley, A., Tata, P., Murray, H., & Palmer, S. (2006). The effectiveness of cognitive behavior therapy for borderline personality disorder: Results from the Borderline Personality Disorder Study of Cognitive Therapy (BOSCOT) trial. Journal of Per- sonality Disorders, 20, 450 – 465. http://dx.doi.org/10.1521/pedi.2006 .20.5.450

�Donaldson, D., Spirito, A., & Esposito-Smythers, C. (2005). Treatment for adolescents following a suicide attempt: Results of a pilot trial. Journal of the American Academy of Child & Adolescent Psychiatry, 44, 113–120. http://dx.doi.org/10.1097/00004583-200502000-00003

Dubicka, B., Elvins, R., Roberts, C., Chick, G., Wilkinson, P., & Goodyer, I. M. (2010). Combined treatment with cognitive– behavioural therapy in adolescent depression: Meta-analysis. The British Journal of Psychiatry, 197, 433– 440. http://dx.doi.org/10.1192/bjp.bp.109.075853

�Esposito-Smythers, C., Spirito, A., Kahler, C. W., Hunt, J., & Monti, P. (2011). Treatment of co-occurring substance abuse and suicidality among adolescents: A randomized trial. Journal of Consulting and Clinical Psychology, 79, 728 –739. http://dx.doi.org/10.1037/a0026074

�Esposito-Smythers, C., Spirito, A., Uth, R., & LaChance, H. (2006). Cognitive behavioral treatment for suicidal alcohol abusing adolescents: Development and pilot testing. The American Journal on Addictions, 15(Suppl. 1), 126 –130. http://dx.doi.org/10.1080/10550490601006188

�Fleischhaker, C., Böhme, R., Sixt, B., Brück, C., Schneider, C., & Schulz, E. (2011). Dialectical behavioral therapy for adolescents (DBT-A): A clinical trial for patients with suicidal and self-injurious behavior and borderline symptoms with a one-year follow-up. Child and Adolescent Psychiatry and Mental Health, 5, 3. http://dx.doi.org/10.1186/1753- 2000-5-3

Fuller-Thomson, E., Hamelin, G. P., & Granger, S. J. R. (2013). Suicidal ideation in a population-based sample of adolescents: Implications for family medicine practice. ISRN Family Medicine, 2013, 1–11. http:// www.hindawi.com/journals/isrn/2013/282378/

�Goodyer, I., Dubicka, B., Wilkinson, P., Kelvin, R., Roberts, C., Byford, S., . . . Harrington, R. (2007). Selective serotonin reuptake inhibitors (SSRIs) and routine specialist care with and without cognitive behaviour therapy in adolescents with major depression: Randomised controlled trial. British Medical Journal, 335, 142–146. http://dx.doi.org/10.1136/ bmj.39224.494340.55

�Green, J. M., Wood, A. J., Kerfoot, M. J., Trainor, G., Roberts, C., Rothwell, J., . . . Harrington, R. (2011). Group therapy for adolescents with repeated self-harm: Randomised controlled trial with economic evaluation. British Medical Journal, 342, d682. http://dx.doi.org/ 10.1136/bmj.d682

Groves, S., Backer, H. S., van den Bosch, W., & Miller, A. (2012). Dialectical behaviour therapy with adolescents. Child and Adolescent Mental Health, 17, 65–75. http://dx.doi.org/10.1111/j.1475-3588.2011 .00611.x

�Harrington, R., Kerfoot, M., Dyer, E., McNiven, F., Gill, J., Harrington, V., . . . Byford, S. (1998). Randomized trial of a home-based family

intervention for children who have deliberately poisoned themselves. Journal of the American Academy of Child & Adolescent Psychiatry, 37, 512–518. http://dx.doi.org/10.1016/S0890-8567(14)60001-0

Hawton, K., Saunders, K. E. A., & O’Connor, R. C. (2012). Self-harm and suicide in adolescents. The Lancet, 379, 2373–2382. http://dx.doi.org/ 10.1016/S0140-6736(12)60322-5

�Hazell, P. L., Martin, G., McGill, K., Kay, T., Wood, A., Trainor, G., & Harrington, R. (2009). Group therapy for repeated deliberate self-harm in adolescents: Failure of replication of a randomized trial. Journal of the American Academy of Child & Adolescent Psychiatry, 48, 662– 670.

Hedges, L., & Olkin, I. (1985). Statistical methods for meta-analysis. New York, NY: Academic Press.

Higgins, J. P. T., & Green, S. (2011). Cochrane handbook for systematic reviews of interventions (Version 5.1.0). The Cochrane Collaboration. Retrieved from http://www.cochrane-handbook.org

�James, A. C., Taylor, A., Winmill, L., & Alfoadari, K. (2008). A prelim- inary community study of dialectical behaviour therapy with adolescent females demonstrating persistent, deliberate self-harm. Child and Ado- lescent Mental Health, 13, 148 –152. http://dx.doi.org/10.1111/j.1475- 3588.2007.00470.x

�James, A. C., Winmill, L., Anderson, C., & Alfoadari, K. (2011). A preliminary study of an extension of a community dialectic behaviour therapy programme to adolescents in the looked after care system. Child and Adolescent Mental Health, 16, 9 –13. http://dx.doi.org/10.1111/j .1475-3588.2010.00571.x

�Katz, L. Y., Cox, B. J., Gunasekara, S., & Miller, A. L. (2004). Feasibility of dialectical behavior therapy for suicidal adolescent inpatients. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 276 – 282. http://dx.doi.org/10.1097/00004583-200403000-00008

Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., . . . Ryan, N. (1997). Schedule for Affective Disorders and Schizophrenia for School-Age Children—Present and Lifetime Version (K-SADS–PL): Initial reliability and validity data. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 980 –988. http://dx.doi.org/10.1097/ 00004583-199707000-00021

Kerfoot, M. (1988). Deliberate self-poisoning in childhood and early adolescence. Child Psychology and Psychiatry, 29, 335–343. http://dx .doi.org/10.1111/j.1469-7610.1988.tb00721.x

Klein, D. A., & Miller, A. L. (2011). Dialectical behavior therapy for suicidal adolescents with borderline personality disorder. Child and Adolescent Psychiatric Clinics of North America, 20, 205–216. http:// dx.doi.org/10.1016/j.chc.2011.01.001

Kutcher, S. P., & Szumilas, M. (2008). Youth suicide prevention. Cana- dian Medical Association Journal, 178, 282–285. http://dx.doi.org/ 10.1503/cmaj.071315

Linehan, M. M., & Comtois, K. A. (1994). Lifetime parasuicide count. Seattle, WA: University of Washington.

Mann, J. J., McBride, P. A., Brown, R. P., Linnoila, M., Leon, A. C., DeMeo, M., . . . Stanley, M. (1992). Relationship between central and peripheral serotonin indexes in depressed and suicidal psychiatric inpa- tients. Archives of General Psychiatry, 49, 442– 446. http://dx.doi.org/ 10.1001/archpsyc.1992.01820060022003

�March, J., Silva, S., Petrycki, S., Curry, J., Wells, K., Fairbank, J., . . . Severe, J. (2004). Fluoxetine, cognitive– behavioral therapy, and their combination for adolescents with depression: Treatment for Adolescents With Depression Study (TADS) randomized controlled trial. Journal of the American Medical Association, 292, 807– 820. http://dx.doi.org/ 10.1001/jama.292.7.807

�McDonell, M. G., Tarantino, J., Dubose, A. P., Matestic, P., Steinmetz, K., Galbreath, H., & McClellan, J. M. (2010). A pilot evaluation of dialectical behavioural therapy in adolescent long-term inpatient care. Child and Adolescent Mental Health, 15, 193–196. http://dx.doi.org/ 10.1111/j.1475-3588.2010.00569.x

377EVIDENCE-BASED PRACTICE

�Melvin, G. A., Tonge, B. J., King, N. J., Heyne, D., Gordon, M. S., & Klimkeit, E. (2006). A comparison of cognitive– behavioral therapy, sertraline, and their combination for adolescent depression. Journal of the American Academy of Child & Adolescent Psychiatry, 45, 1151– 1161. http://dx.doi.org/10.1097/01.chi.0000233157.21925.71

Muehlenkamp, J. J., Claes, L., Havertape, L., & Plener, P. L. (2012). International prevalence of adolescent non-suicidal self-injury and de- liberate self-harm. Child and Adolescent Psychiatry and Mental Health, 6, 10. http://dx.doi.org/10.1186/1753-2000-6-10

O’Connor, R. C., & Nock, M. K. (2014). The psychology of suicidal behaviour. The Lancet, 1, 73– 85.

�Oldershaw, A., Simic, M., Grima, E., Jollant, F., Richards, C., Taylor, L., & Schmidt, U. (2012). The effect of cognitive behavior therapy on decision making in adolescents who self-harm: A pilot study. Suicide and Life-Threatening Behavior, 42, 255–265. http://dx.doi.org/10.1111/ j.1943-278X.2012.0087.x

Orwin, R. G. (1983). A fail-safe N for effect size in meta-analysis. Journal of Educational Statistics, 8, 157–159. http://dx.doi.org/10.2307/1164923

Ougrin, D., Tranah, T., Leigh, E., Taylor, L., & Asarnow, J. R. (2012). Practitioner review: Self-harm in adolescents. Journal of Child Psychol- ogy and Psychiatry, 53, 337–350. http://dx.doi.org/10.1111/j.1469-7610 .2012.02525.x

Pelkonen, M., & Marttunen, M. (2003). Child and adolescent suicide: Epidemiology, risk factors, and approaches to prevention. Paediatric Drugs, 5, 243–265. http://dx.doi.org/10.2165/00128072-200305040- 00004

Perlman, C. M., Neufeld, E., Martin, L., Goy, M., & Hirdes, J. P. (2011). Suicide risk assessment inventory: A resource guide for Canadian health care organizations. Toronto, ON: Canadian Patient Safety Institute and Ontario Hospital Association.

Poznanski, E. O., & Mokros, H. B. (1995). Children’s Depression Rating Scale—Revised (CDRS–R). Los Angeles, CA: Western Psychological Services.

�Rathus, J. H., & Miller, A. L. (2002). Dialectical behavior therapy adapted for suicidal adolescents. Suicide and Life-Threatening Behavior, 32, 146 –157. http://dx.doi.org/10.1521/suli.32.2.146.24399

Reynolds, W. M. (1988). Suicidal Ideation Questionnaire, professional manual. Odessa, FL: Psychological Assessment Resources.

Reynolds, W. M., & Mazza, J. J. (1999). Assessment of suicidal ideation in inner-city children and young adolescents. School Psychology Review, 28, 17–30.

Robinson, J., Hetrick, S. E., & Martin, C. (2011). Preventing suicide in young people: Systematic review. Australian and New Zealand Journal of Psychiatry, 45, 3–26. http://dx.doi.org/10.3109/00048674.2010 .511147

�Rohde, P., Clarke, G. N., Mace, D. E., Jorgensen, J. S., & Seeley, J. R. (2004). An efficacy/effectiveness study of cognitive– behavioral treat- ment for adolescents with comorbid major depression and conduct disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 660 – 668. http://dx.doi.org/10.1097/01.chi.0000121067 .29744.41

Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods and data analysis. Montreal, QC: McGraw-Hill.

Rosenthal, R., & Rubin, D. B. (1982a). Comparing effect sizes of inde- pendent studies. Psychological Bulletin, 92, 500 –504. http://dx.doi.org/ 10.1037/0033-2909.92.2.500

Rosenthal, R., & Rubin, D. B. (1982b). Further meta-analytic procedures assessing cognitive gender differences. Journal of Educational Psychol- ogy, 74, 708 –712. http://dx.doi.org/10.1037/0022-0663.74.5.708

Spirito, A., Esposito-Smythers, C., Wolff, J., & Uhl, K. (2011). Cognitive– behavioral therapy for adolescent depression and suicidality. Child and Adolescent Psychiatric Clinics of North America, 20, 191–204. http:// dx.doi.org/10.1016/j.chc.2011.01.012

Statistics Canada. (2014). Suicides and suicide rate, by sex and by age group (CANSIM: Table 102– 0551). Retrieved from http://www.statcan .gc.ca/tables-tableaux/sum-som/l01/cst01/hlth66a-eng.htm

Tarrier, N., Taylor, K., & Gooding, P. (2008). Cognitive– behavioral in- terventions to reduce suicide behavior: A systematic review and meta- analysis. Behavior Modification, 32, 77–108. http://dx.doi.org/10.1177/ 0145445507304728

Tarrier, N., & Wykes, T. (2004). Is there evidence that cognitive behaviour therapy is an effective treatment for schizophrenia? A cautious or cautionary tale? Behaviour Research and Therapy, 42, 1377–1401. http://dx.doi.org/10.1016/j.brat.2004.06.020

�Taylor, L. M. W., Oldershaw, A., Richards, C., Davidson, K., Schmidt, U., & Simic, M. (2011). Development and pilot evaluation of a manu- alized cognitive– behavioural treatment package for adolescent self- harm. Behavioural and Cognitive Psychotherapy, 39, 619 – 625. http:// dx.doi.org/10.1017/S1352465811000075

Townsend, E., Walker, D. M., Sargeant, S., Vostanis, P., Hawton, K., Stocker, O., & Sithole, J. (2010). Systematic review and meta-analysis of interventions relevant for young offenders with mood disorders, anxiety disorders, or self-harm. Journal of Adolescence, 33, 9 –20. http://dx.doi.org/10.1016/j.adolescence.2009.05.015

�Vitiello, B., Brent, D. A., Greenhill, L. L., Emslie, G., Wells, K., Walkup, J. T., . . . Zelazny, J. (2009). Depressive symptoms and clinical status during the Treatment of Adolescent Suicide Attempters (TASA) study. Journal of the American Academy of Child & Adolescent Psychiatry, 48, 997–1004.

Wagner, B. M. (2009). Suicidal behaviour in children and adolescents. New Haven, CT: Yale University Press.

Wasserman, D., Rihmer, Z., Rujescu, D., Sarchiapone, M., Sokolowski, M., Titelman, D., . . . Carli, V. (2012). The European Psychiatric Association (EPA) guidance on suicide treatment and prevention. Eu- ropean Psychiatry, 27, 129 –141. http://dx.doi.org/10.1016/j.eurpsy .2011.06.003

Weisz, J. R., McCarty, C. A., & Valeri, S. M. (2006). Effects of psycho- therapy for depression in children and adolescents: A meta-analysis. Psychological Bulletin, 132, 132–149. http://dx.doi.org/10.1037/0033- 2909.132.1.132

Wolf, F. M. (1986). Meta-analysis: Quantitative methods for research synthesis. Newbury Park, CA: Sage.

�Wood, A., Trainor, G., Rothwell, J., Moore, A., & Harrington, R. (2001). Randomized trial of group therapy for repeated deliberate self-harm in adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 1246 –1253. http://dx.doi.org/10.1097/00004583- 200111000-00003

�Woodberry, K. A., & Popenoe, E. J. (2008). Implementing dialectical behavior therapy with adolescents and their families in a community outpatient clinic. Cognitive and Behavioral Practice, 15, 277–286. http://dx.doi.org/10.1016/j.cbpra.2007.08.004

Received July 1, 2014 Revision received March 2, 2015

Accepted March 3, 2015 �

378 LABELLE, POULIOT, AND JANELLE