NEW RESEARCH

Dialectical Behavior Therapy for Adolescents

With Repeated Suicidal and Self-harming
Behavior: A Randomized Trial
Lars Mehlum, MD, PhD, Anita J. Tørmoen, MA, Maria Ramberg, MD, Egil Haga, PhD,
Lien M. Diep, MSc, Stine Laberg, MA, Bo S. Larsson, MD, PhD, Barbara H. Stanley, PhD,
Alec L. Miller, PsyD, Anne M. Sund, MD, PhD, Berit Grøholt, MD, PhD

Objective: We examined whether a shortened form of dialectical behavior therapy, dia-

lectical behavior therapy for adolescents (DBT-A) is more effective than enhanced usual
care (EUC) to reduce self-harm in adolescents. Method: This was a randomized study of
77 adolescents with recent and repetitive self-harm treated at community child and adolescent
psychiatric outpatient clinics who were randomly allocated to either DBT-A or EUC. Assessments
of self-harm, suicidal ideation, depression, hopelessness, and symptoms of borderline personality
disorder were made at baseline and after 9, 15, and 19 weeks (end of trial period), and frequency
of hospitalizations and emergency department visits over the trial period were record-
ed. Results: Treatment retention was generally good in both treatment conditions, and the use
of emergency services was low. DBT-A was superior to EUC in reducing self-harm, suicidal
ideation, and depressive symptoms. Effect sizes were large for treatment outcomes in patients
who received DBT-A, whereas effect sizes were small for outcomes in patients receiving EUC.
Total number of treatment contacts was found to be a partial mediator of the association between
treatment and changes in the severity of suicidal ideation, whereas no mediation effects were
found on the other outcomes or for total treatment time. Conclusion: DBT-A may be an
effective intervention to reduce self-harm, suicidal ideation, and depression in adolescents with
repetitive self-harming behavior. Clinical trial registration information—Treatment for Adoles-
cents With Deliberate Self Harm; http://ClinicalTrials.gov/; NCT00675129. J. Am. Acad. Child
Adolesc. Psychiatry, 2014;53(10):1082–1091. Key Words: self-harm, attempted suicide, psy-
chotherapy, randomized trial

S
elf-harming behavior (nonfatal self- reasons for such behavior.2,3 Only 10% to 20%
poisoning or self-injury with or without of adolescents who have self-harmed report
suicide intent)1 in adolescents is a serious receiving any kind of subsequent treatment.4
public health problem in many countries. Ac- Even fewer report having received treatment in
cording to population studies, between 5% and child and adolescent psychiatric services. Among
10% of adolescents report past-year self-harm, those who are referred to specialized care, many
with cutting as the most commonly reported will be noncompliant with the treatment or will
method. Relief from intensely unpleasant emo- drop out prematurely because of treatment-
tions or dying are reported as the most common interfering characteristics of the adolescents,
their families, or the clinical services. Self-harm
is a powerful predictor of completed suicide.5
There is thus a strong need to develop effective
This article is discussed in an editorial by Dr. Alan Apter on
page 1048.
interventions that are accessible and acceptable
to adolescents and their families, as well as
Clinical guidance is available at the end of this article. feasible for clinicians in community mental
This article can be used to obtain continuing medical education health settings.
(CME) at www.jaacap.org. Repetitive self-harm is very often linked to
personality disorders, in particular to borderline

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

1082 www.jaacap.org VOLUME 53 NUMBER 10 OCTOBER 2014
 EFFICACY OF DBT FOR SUICIDAL ADOLESCENTS

personality disorder (BPD); this has been months to 3 to 5 months, including parents or
well demonstrated in adult populations.6 Affective other caregivers in weekly skills training groups,
instability and a pronounced sensitivity to envi- and adding a new skills module to address
ronmental stress are among the BPD characteris- common skill deficits among teens with emotion
tics shown to increase vulnerability for suicidal dysregulation and their families. Several un-
and self-harm behavior.7 Individuals diagnosed controlled studies have suggested that DBT-A
with BPD are typically highly emotionally reac- could be effective at reducing self-harm while
tive, their reactions tend to be extreme, and the improving treatment compliance and satisfac-
time taken to return to their baseline affective state tion.22-27 However, to date, no RCTs of DBT-A
is often considerably longer than for individuals have been published. Although RCT studies of
without BPD. Although clinicians and researchers standard DBT for adults offer favorable results,
diagnose personality disorders primarily in adults, our knowledge of the effectiveness of a consid-
in the past decade, more clinical researchers have erably shorter and modified DBT for adolescents
assessed and identified personality traits and dis- and their families is limited. The primary hy-
orders in youth.8 Such traits are highly predictive pothesis consequently examined in this study
of adult personality disorders and are associated was that DBT-A would be superior to usual care
with increased long-term impairment, morbidity, in reducing self-harm behavior, suicidal ideation,
and mortality.9 An important question to answer is and depressive symptoms in self-harming ado-
whether therapeutic interventions in adolescence lescents with BPD features.
could prevent the development of adult personal-
ity disorders linked to self-harming behavior.
The paucity of randomized controlled trials
METHOD
This was a single-blind randomized trial comparing
(RCTs) offers scarce evidence for effective treat- DBT-A with enhanced usual care (EUC). Participants
ments for self-harming adolescents. Wood et al.10 were randomly allocated (Figure 1) to receive either
developed a manual-based cognitive-behavioral treatment at 1 of the participating child and adolescent
therapy–oriented group therapy for adolescents psychiatric outpatient clinics in a 1:1 ratio stratified
that was shown to be more effective than treat- according to gender, presence of major depression, and
ment as usual (TAU) in reducing self-harm presence of suicide intent during the most serious
behaviors; however, these results were not repli- episode of self-harm behavior within the 16 weeks
cated in 2 subsequent studies from the same before enrollment. Treatment allocation of participants
group.11,12 In a recent RCT reported by Rossouw after baseline assessments was based on a permuted
block randomization procedure with an undisclosed
and Fonagy,13 mentalization-based treatment for
and variable blocking factor, and daily management of
adolescents was superior to TAU in reducing self-
the randomization procedures was performed by an
harm and depression. However, with this notable external group.
exception, no treatment program specifically tar-
geting self-harm in adolescents has so far been
Participants
shown to reduce self-harm more than usual care.
Participants were 77 adolescents (age 12 through 18
The recent critical review of Brent et al.14 provides
years) recruited from child and adolescent psychiatric
a good update on intervention studies targeting outpatient clinics in Oslo that screened newly referred
adolescent suicidal behaviors. patients for current self-harm behavior. If screened
Dialectical behavior therapy (DBT) is a positively, the patient and the parents were invited to a
comprehensive, principle-based, multi-modal, diagnostic interview in which the remaining inclusion
outpatient treatment that was developed by criteria were checked. Diagnostic assessments were
Linehan6 for adults with BPD; it was found, in made by experienced clinicians blinded to treatment
several RCTs at independent sites, to be superior allocation. Inclusion criteria were as follows: a history
to comparison treatments in reducing suicidal of at least 2 episodes of self-harm, at least 1 within
and nonsuicidal self-harm, emergency depart- the last 16 weeks; at least 2 criteria of DSM-IV BPD
(plus the self-destructive criterion), or, alternatively,
ment visits, and hospitalizations, improving out-
at least 1 criterion of DSM-IV BPD plus at least
patient treatment completion, global and social
2 subthreshold-level criteria; and fluency in Norwe-
adjustment, and personality functioning.15-20 DBT gian. Exclusion criteria were a diagnosis of bipolar
has since been adapted by Miller et al. for out- disorder (except bipolar II), schizophrenia, schizo-
patient treatment of self-harming adolescents affective disorder, psychotic disorder not otherwise
(DBT-A) with borderline personality traits,21 specified, intellectual disability, and Asperger syn-
through shortening treatment length from 12 drome. Self-harm was defined as self-poisoning or

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 53 NUMBER 10 OCTOBER 2014 www.jaacap.org 1083
MEHLUM et al.

FIGURE 1 CONsolidated Standards of Reporting Trials (CONSORT) flowchart of participants, comparing dialectical
behavior therapy adapted for adolescents (DBT-A) with enhanced usual care (EUC) for suicidal and self-harming
behavior. Note: BPD ¼ borderline personality disorder.

self-injury irrespective of intent,1 including self-harm Dialectical Behavior Therapy. Dialectical Behavior
with suicidal intent, nonsuicidal self-harm, and self- Therapy,21 delivered for 19 weeks, consisted of
harm episodes with unclear intent. From March 2008 1 weekly session of individual therapy (60 minutes),
to March 2012, a total of 152 screen-positive adoles- 1 weekly session of multifamily skills training (120
cents recruited from the participating child and minutes), and family therapy sessions and telephone
adolescent psychiatric outpatient clinics plus an addi- coaching with individual therapists outside therapy
tional 142 adolescents referred directly to the study sessions as needed. Fifteen psychologists and psychia-
from general practitioners, child protection services, or trists previously unfamiliar with DBT were recruited
school health services were initially checked for inclu- for the purpose of the trial and were trained through an
sion criteria (Figure 1). Of these, 97 were assessed 80-hour seminar with an additional 12 months of su-
through interviews, and of these, 77 adolescents were pervised practice on clinical training cases, and were
included in the study and randomized. rated for adherence to DBT-A treatment principles.
From these candidates, 8 therapists (2 psychiatrists,
5 clinical psychologists, and 1 educational psycholo-
Treatments and Therapists gist) were selected after having completed a consis-
Patients received either DBT or EUC by therapists tently adherent training case, that is, a patient for
working at and funded by the 10 child and adolescent whom the therapist under training (in DBT-A) was able
psychiatric outpatient clinics participating in the study. to deliver the treatment consistently at the adherent
These clinics were all publicly funded, and all treat- level (score of 4.0 or above) throughout the 19 weeks of
ments were provided free of charge to the patients and treatment of that patient. Adherence to DBT continued
their families. Therapists provided only DBT or EUC. to be assessed throughout the trial. For each patient–
Patients received ancillary non-manualized pharma- therapist dyad in individual therapy, 5 videotaped
cotherapy as needed. sessions (first 2 sessions and 3 random) were rated by

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

1084 www.jaacap.org VOLUME 53 NUMBER 10 OCTOBER 2014
 EFFICACY OF DBT FOR SUICIDAL ADOLESCENTS

an independent rater (S.L.) trained to and maintaining (LPC) interview33 was used to obtain self-harm history,
reliability with the treatment developer group in the and the Suicide Intent Scale (SIS)34 evaluated suicide
use of the DBT Global Rating Scale (M.M. Linehan, intent at the most severe self-harm episode during the
unpublished, 2003), a 64-item instrument scored from last 4 months. Participants received a small amount of
0 to 5, with higher scores reflecting higher adherence. monetary compensation for each assessment session.
On average, 1 randomly selected videotaped skills
training session per group was rated per month.
Adherence was defined as a sum score of 4.0. Outcome Measures
Enhanced Usual Care. Enhanced usual care was 19 The primary outcomes were as follows: the number of
weeks of standard care (enhanced for the purpose of self-reported self-harm episodes (suicide attempts and
the study by requiring that EUC therapists agree to non-suicidal self-harm episodes combined); the
provide on average no less than 1 weekly treatment severity of suicidal ideation as measured by the 15-item
session per patient throughout the trial) delivered by self-report Suicidal Ideation Questionnaire (SIQ-JR;
therapists (4 psychiatrists, 16 clinical psychologists, suicidal thoughts rated on a 7-point scale from “I never
6 clinical social workers, 2 clinical pedagogues, had this thought” to “almost every day”)35; and level of
1 specialist nurse, and 1 psychology graduate student) depressive symptoms as measured by the 13-item
not trained in or practicing DBT. EUC was not man- version of the self-report Short Mood and Feelings
ualized or checked for fidelity, and was either Questionnaire (SMFQ)36 and through the interviewer-
psychodynamically oriented therapy or cognitive- rated 10-item Montgomery–Asberg Depression Rating
37
behavioral therapy combined with psychopharmaco- Scale (MADRS) during the treatment trial period.
logical treatment as needed. EUC was delivered for a Other outcomes were hopelessness, measured by the
minimum of 19 weeks but could extend beyond the 20-item self-report Beck Hopelessness Scale (BHS)38;
trial time window, depending on the EUC therapists’ borderline symptoms, assessed through the 23-item
assessments of their patients’ needs. (Further details are self-report Borderline Symptom List (BSL)39; and hos-
provided in the section on treatment implementation pital admissions and emergency department visits
and retention). because of self-harm during the trial. The SIQ-JR,
Patient Safety. The study complied with National SMFQ, BHS, and BSL were measured at baseline and
Institute of Mental Health (NIMH) recommendations28 after 9, 15, and 19 weeks, whereas self-harm was
for intervention research with patients at high risk for measured at 9 weeks (covering the first 9 weeks) and 15
suicidality. All study therapists received suicide risk weeks (covering the next 6 weeks), and MADRS was
assessment and management training before patient measured at baseline and 19 weeks.
treatment commenced. For both treatment modalities,
results from the baseline assessments of suicide and
self-harm risk, psychiatric diagnoses, and symptom Raters and Integrity of Ratings
severity were made available to the attending thera- Two child and adolescent psychiatrists and 2 doctoral-
pists before the first therapy session. Also, when a level clinicians, blinded to treatment allocation, con-
patient’s follow-up data indicated high risk of self- ducted the baseline interviews. Ten independent
harm or suicide, the study management immediately assessors, blinded to treatment allocation and to results
notified the patient’s therapist. The study was from baseline interviews, conducted interviews at trial
approved by the Regional Committee for Medical completion. To ensure the integrity of blinding, a
Research Ethics, South-East Norway, and all patients nonblinded project coordinator made all of the prac-
and parents provided written informed consent. tical arrangements for follow-up interviews and
collected treatment history data. All patients were
instructed not to disclose any information about their
Assessments treatment. When asked after completion of interviews
Assessments were performed by independent in- which treatment they thought each patient received,
terviewers blind to treatment allocation at baseline assessors’ responses were correct for 44.2% of patients
(interview and self-report) before randomization and (Cohen’s k ¼ 0.12), indicating that blinding was suc-
then at 9 (self-report), 15 (self-report), and 19 weeks cessful. All interviews were audio-taped, and interrater
(interview and self-report) after the first treatment reliability (IRR) of diagnoses and outcome variables
session. The Schedule for Affective Disorders–Present was checked by a child and adolescent psychiatrist
and Lifetime version (K-SADS-PL)29 was used to obtain (A.M.S.) who was expert in the relevant assessment
demographics and DSM-IV Axis I diagnoses and the instruments. Based on 26 IRR-rated interviews, the
Structured Clinical Interview for DSM-IV (SCID-II)30 mean k value was 0.68 (range ¼ 0.50–0.81, standard
was used to evaluate BPD. The Child Behavior deviation [SD] ¼ 0.10) for all symptoms rated with K-
Checklist (CBCL)31 for parents measured adolescents’ SADS-PL. Intraclass correlation (ICC) was used to test
emotional and behavioral problems, and the Children’s IRR for C-GAS (ICC ¼ 0.42), MADRS score (ICC ¼
Global Assessment Scale (C-GAS)32 measured global 0.76), LPC (IRR ¼ 0.99), and SCID-II diagnostic criteria
level of impairment. The Lifetime Parasuicide Count for BPD (ICC ¼ 0.66).

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 53 NUMBER 10 OCTOBER 2014 www.jaacap.org 1085
MEHLUM et al.

Statistical Analysis Pocock,40 conducted after completion of the first 40

As exact data on the repetition rate of self-harm over a patients where no research dropouts had been
19-week observation period for adolescents treated observed, and with a self-harm repetition of 30% for
with DBT-A or EUC were not available when the study DBT patients and 83% for EUC patients, revealed that
was planned, the power analysis anticipated a repeti- the necessary number of patients to be included could
tion rate over this time span of 50% for adolescents be reduced to 80 (40 in each group).
receiving EUC, and 25% for patients receiving DBT-A Data analysis was by intention to treat. Means and
(building on adult studies of DBT). The power anal- standard deviations or median and interquartile ranges
ysis anticipated a 25% research assessment drop-out were computed for normally and non-normally
rate. With an a error level of 0.05, a sample of 150 distributed variables. Differences between central ten-
patients (75 in each group) would be required to pro- dencies in the groups were tested by independent-
vide 80% power with a 2-tailed test. To limit the sample t tests or Mann–Whitney U tests. Differences
number of patients subjected to potential nonresponse, between the group proportions were tested by Pearson
a revised mid-study power analysis was planned. This c2 or Fisher exact tests. Estimation of trend and dif-
analysis, using a more stringent nominal significance ferences between group trends over time were exam-
level (0.029) according to the recommendations by ined by mixed-effects multiple regression, with the sum

TABLE 1 Baseline Demographic and Diagnostic Data and Number of Suicide Attempts and Nonsuicidal Self-Harm
Episodes in Adolescent Patients (N ¼ 77) Allocated to Receive 19 Weeks of Dialectical Behavior Therapy (DBT) or
Enhanced Usual Care (EUC)
Total Sample
DBT (n ¼ 39) EUC (n ¼ 38) (N ¼ 77)

Variablea n %b n %b N %b
Female sex 34 87.2 34 89.5 68 88.3
Norwegian ethnicity 30 78.9 32 91.4 62 84.9
High school graduate 15 39.5 12 32.4 27 36.0
Parents currently married 17 43.6 17 44.7 34 44.2
Child protection (current) 6 15.4 7 18.4 13 16.9
Child protection (past) 10 26.3 11 28.9 21 27.6
Past psychiatric treatment 28 73.7 23 62.2 51 68.0
Past psychopharmacotherapy 2 5.4 6 17.1 8 11.1
Current psychopharmacotherapy 6 15.4 3 7.9 9 11.7
Current DSM-IV Axis I and II diagnoses
MDD 9 23.1 8 21.1 17 22.1
Other depressive disorder 16 41.0 13 34.2 29 37.7
Panic disorder 2 5.1 5 13.2 7 9.1
PTSD 7 17.9 6 15.8 13 16.9
Any anxiety disorder 18 46.2 15 39.5 33 42.9
Any SUD 1 2.6 1 2.6 2 2.6
Any eating disorder 3 7.7 3 7.9 6 7.8
BPD 10 26.3 5 14.3 15 20.5
Attempted suicide last 4 months 11 28.2 9 23.7 20 26.0

Mean SD Mean SD Mean SD

Age (y) 15.9 1.4 15.3 1.6 15.6 1.5
C-GAS score 55.3 8.0 57.9 10.1 56.1 8.3
CBCL total score, by parent (n) 69.6 11.0 68.4 8.9 69.0 9.8
BPD criteria fulfilled (n)c 4.0 2.0 3.0 3.0 4.0 2.0
Current DSM-IV Axis I disorders (n)c 2.0 1.0 1.0 3.0 1.0 2.0
Suicide attempts, lifetime (n)d 2.1 5.2 1.3 2.8 1.7 4.2
Nonsuicidal self-harm, lifetime (n)c 49.5 159.5 25.0 45.5 34.0 88.0
Note: BPD ¼ borderline personality disorder; CBCL ¼ Child Behavior Checklist; C-GAS ¼ Children’s Global Assessment Scale; MDD ¼ major depressive
disorder; PTSD ¼ posttraumatic stress disorder; SUD ¼ substance use disorder.
a
No significant differences between the treatment groups were found on any baseline characteristics.
b
There were slight variations in the percentage basis because of missing data in some cells.
c
Median and interquartile range.
d
The median was 0 for both groups. The interquartile ranges were 1.0 and 1.3 in the DBT and EUC groups, respectively.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

1086 www.jaacap.org VOLUME 53 NUMBER 10 OCTOBER 2014
 EFFICACY OF DBT FOR SUICIDAL ADOLESCENTS

scores as dependent variables. A random intercept and any treatment during this 19-week period. Within
slope for each patient with an unstructured variance– this period, DBT-A patients attended an average
covariance matrix was used. A likelihood ratio test was of 13.8 (SD ¼ 6.9) individual sessions, 11.2 (SD ¼
used to examine whether the time trend was nonlinear 5.9) skills training group sessions, 2.6 (SD ¼ 2.2)
using the maximum likelihood method. For data on
family therapy sessions, and 3.3 (SD ¼ 4.5) brief
repeated self-harm over the course of the trial, gener-
intersession telephone contacts. The correspond-
alized estimating equations (GEE) with Poisson and
robust variance were used to test for between-group ing figures for EUC patients were 11.5 (SD ¼ 6.4)
differences in the average numbers of self-harm. individual sessions, 0.5 (SD ¼ 2.1) group therapy
Mediation analyses were conducted using structural sessions, 5.8 (SD ¼ 9.8) family therapy sessions,
equation models for outcomes with normal distribu- and 3.5 (SD ¼ 4.4) telephone sessions. Except for
tion and generalized structural equation models for group sessions, there were no significant
count data with mean overdispersion. The maximum between-group differences in the number of ses-
likelihood estimation method was used in all media- sions attended. The average number of weeks
tion analyses. All tests were 2-sided, and the signifi- (holiday absence omitted) receiving any treat-
cance level was set to .05. Analyses were performed ment was 14.9 (SD ¼ 5.6) for DBT-A patients and
with STATA 1341 and IBM SPSS Statistics 20.0 for
12.9 (SD ¼ 5.1) for EUC patients. More than 3
Windows.42
dropped individual therapy sessions is consid-
ered dropout in DBT-A, and according to this
RESULTS criterion, 10 patients (25.6%) dropped out. In the
EUC group, where such strict dropout rules were
Sample Characteristics
not applied, 11 patients (28.9%) attended less
Screening, assessment, and randomization pro-
than 50% of the weekly individual therapy ses-
cedures leading to allocation of 77 adolescents
sions. Three DBT-A patients (7.7%) used at least
to either DBT-A (n ¼ 39) or EUC (n ¼ 38)
1 psychotropic drug for a mean number of 94.7
are summarized in Figure 1. Although some pa-
days (SD ¼ 64.3), whereas 5 EUC patients (13.2%)
tients dropped out of treatment, all patients
used such medication for a mean number of 72.8
were followed from baseline to trial completion
days (SD ¼ 16.6), with no significant differences.
with no dropouts from the research. Baseline
demographic characteristics, diagnostic variables,
and pretreatment suicidal and nonsuicidal self-
harm behaviors are displayed in Table 1. No sig- FIGURE 2 Comparison between patients receiving
nificant differences between the treatment groups dialectical behavior therapy adapted for adolescents
were found on any baseline characteristics. (DBT-A; n ¼ 39) and enhanced usual care (EUC; n ¼ 38)
in frequency of self-harming episodes during the
treatment trial. Note: Frequency of self-harm episodes
Treatment Implementation and Retention was compared for the time periods from baseline to week
Altogether, 169 individual DBT sessions from the 9 of the trial and from week 10 to week 15. Generalized
39 DBT patient–therapist dyads and 56 multi- estimating equation analyses with Poisson distribution
family skills-training group sessions were scored using exchangeable correlation structure with robust
for adherence. The mean score for individual variance showed that only the DBT-A group significantly
therapy sessions was 4.11 (SD ¼ 0.14; range ¼ 3.43– decreased self-harm frequency, and the between-group
4.37) and for multifamily skills-training sessions, difference was statistically significant (D slope¼ 0.92,
95% CI ¼ 1.69, 0.15, p ¼ .021).
4.18 (SD ¼ 0.10; range ¼ 3.93–4.36), both within the
adherent range.
Patients in the DBT-A condition were enrolled
in the skills training groups after they had
attended on average 3 individual therapy ses-
sions (according to the DBT-A protocol, new pa-
tients were admitted only at the start of each
4-week skills training module) for practical rea-
sons. The total duration of the treatment was
therefore 19 weeks (3 þ 16). For comparability
reasons, treatment retention in both DBT-A and
EUC patients was counted as number of treat-
ment sessions and number of weeks receiving

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 53 NUMBER 10 OCTOBER 2014 www.jaacap.org 1087
MEHLUM et al.

Suicidal and Self-harm Behaviors and Suicidal regression analysis (D slope ¼ 0.62 per week,
Ideation p ¼ .010).
There were no suicides during the trial period.
The self-harm frequency in the 2 treatment con- Depressive Symptoms
ditions during the 2 measured time intervals of Both patient groups displayed a significant re-
the trial period is shown in Figure 2. Whereas duction in self-reported symptoms of depression,
EUC patients reported a mean of 4.7 (SD ¼ 5.5) but only the DBT-A group showed a significant
episodes in the first 9 weeks and 3.3 (SD ¼ 6.8) reduction in interviewer-rated depression; this
in the subsequent 6 weeks, DBT-A patients re- between-group difference was statistically sig-
ported a mean of 4.1 (SD ¼ 5.8) and 1.2 (SD ¼ nificant according to mixed-effects linear regres-
2.0) episodes in the corresponding time intervals. sion analysis (Table 2).
The average drop on logarithmic scale in self-
harm frequency in the DBT-A group (slope ¼ Outcomes in Patients Who Completed the
1.28, 95% CI ¼ 1.77 to 0.80, p < .001) was Treatment
highly significant, whereas the drop in the A separate series of analyses was conducted with
EUC group (slope ¼ 0.36, 95% CI ¼ 0.99 to only those patients who had dropped less than 4
0.26, p ¼ .254) was not. The between-group treatment sessions (dropout criterion in DBT-A;
difference was statistically significant (D n ¼ 47). The differences between the 2 treat-
slope¼ 0.92, 95% CI ¼ 1.69 to 0.15, p ¼ ment conditions remained significant for all 3
.021). Both treatment groups had a baseline primary outcome variables.
severity of suicidal ideation well above the
clinical cut-off (usually regarded as 31)35 and Associations Between Outcomes and Intensity of
showed similar reductions on this dimension Treatment
over the first 15 weeks (Figure 3 and Table 2), The differences both in the mean duration of the
continuing to drop in DBT-A patients through- treatment and the mean total number of treat-
out the whole trial period, while leveling ment sessions between the 2 treatment condi-
off toward the end of the trial in EUC patients. tions led us to conduct a series of analyses to
This intergroup difference was statistically test whether these 2 variables would mediate
significant according to mixed-effects linear the association between the treatment and the
3 primary outcomes. Total number of treatment
FIGURE 3 Comparison between patients receiving contacts was found to be a partial mediator of
dialectical behavior therapy adapted for adolescents the association between treatment and SIQ-Jr
(DBT-A; n ¼ 39) and enhanced usual care (EUC; n ¼ 38) score, whereas no mediation effects were
in severity of suicidal ideation during the treatment trial. found on the other outcomes or for total treat-
Note: SIQ-Jr ¼ Suicidal Ideation Questionnaire–Junior ment time.
Version. Suicidal ideation was measured at baseline and
after 9, 15, and 19 weeks during the trial. Mixed-effects Other Outcomes
linear regression analysis revealed a statistically The DBT-A group reported reductions in hope-
significant between-group difference in the reduction of
lessness and borderline symptoms (Table 2),
suicidal ideation over time (D slope ¼ 0.62 per week,
but with no significant between-group differ-
p ¼ .010).
ences on these outcomes. One DBT-A patient
and 2 EUC patients were admitted to the hos-
pital because of self-harm during the trial. An
additional 2 DBT-A patients and 5 EUC patients
had an emergency department visit for self-
harm. These differences were not statistically
significant.

DISCUSSION
This was the first randomized trial of the efficacy
of a short version of DBT adapted for the treat-
ment of multi-problem self-harming adolescents
with BPD features. Treatment retention in this

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

1088 www.jaacap.org VOLUME 53 NUMBER 10 OCTOBER 2014
 EFFICACY OF DBT FOR SUICIDAL ADOLESCENTS

TABLE 2 Outcomes for Adolescent Patients (N ¼ 77) Allocated to Receive 19 Weeks of Dialectical Behavior Therapy
(DBT) or Enhanced Usual Care (EUC), by Treatment Assignment
DBT (n ¼ 39) EUC (n ¼ 38)
p Value Effect p Value Effect p-Value
Variable Mean SD Slope Sizea Mean SD Slope Sizea D Slopeb D Slope
Suicidal ideation (SIQ-Jr)
Baseline 36.91 20.82 <.001 0.89 36.91 26.73 .101 0.16 0.62 .010
9 wk 30.72 17.53 31.71 21.20
15 wk 26.58 13.55 28.53 20.97
19 wk 18.30 11.11 32.56 23.99
Hopelessness (BHS)
Baseline 11.48 5.41 <.001 0.97 10.35 5.81 .072 0.22 0.13 .071
9 wk 10.43 5.80 10.00 6.13
15 wk 10.33 5.88 8.48 6.20
19 wk 6.23 5.30 9.06 6.53
Depression (SMFQ)
Baseline 14.92 5.35 <.001 0.88 15.11 6.23 .012 0.41 0.10 .179
9 wk 12.87 6.17 13.39 5.60
15 wk 12.77 6.31 12.23 5.85
19 wk 10.19 5.04 12.58 6.62
Depression (MADRS)
Baseline 19.03 7.84 <.001 0.86 17.50 7.13 .075 0.24 0.22 .019
19 wk 12.29 7.52 15.76 8.14
Borderline symptoms (BSL)
Baseline 38.47 19.32 <.001 0.89 40.18 21.66 .061 0.25 0.50 .050
9 wk 33.63 17.35 34.94 18.57
15 wk 33.85 20.04 33.67 21.28
19 wk 21.34 14.38 34.75 22.15
Note: BHS ¼ Beck Hopelessness Scale; BSL ¼ Borderline Symptoms List; MADRS ¼ Montgomerye Asberg Depression Scale; SIQ-Jr ¼ Suicidal Ideation
QuestionnaireeJunior Version; SMFQ ¼ self-report Mood and Feelings Questionnaire.
a
The mean improvement from baseline to 19 weeks divided by the standard deviation at baseline. All effect sizes are given as positive values.
b
Estimates of difference in slope (per week).

study was generally good, with no differences although there was a nonsignificant trend for
between the 2 treatment conditions. DBT was su- DBT-A patients to have fewer visits.
perior to EUC in reducing frequency of self-harm, Despite the high population prevalence of self-
severity of suicidal ideation, and depressive harm in adolescents, little evidence of effective
symptoms, with generally large effect sizes for treatments exists. Adolescents with borderline
outcomes in the DBT-A condition, but weak or traits and repetitive self-harm have usually been
moderate outcomes in the EUC condition. Inter- considered a difficult-to-treat patient population
estingly, for several outcome measures, differences because of their reluctance to accept treatment
between the treatment conditions increased to- and their tendency to drop out early, and because
ward the last third of the trial period; DBT-A pa- of patient safety issues. This first randomized trial
tients continued their improvement, whereas EUC of DBT-A suggests that it is indeed possible for
patients did not. adolescents to be engaged, retained, and treated
Previous DBT studies in adults have suggested successfully and safely. Only mentalization-based
that teaching specific behavioral skills may be an treatment (MBT-A)13 has so far demonstrated
important factor in the reduction of self-harm and comparable results. From a pragmatic and eco-
depression.43 Early monitoring and explicit tar- nomic perspective, an important difference be-
geting of self-harm throughout the 19 weeks of tween these treatments is the 3-times-longer
DBT-A may also have represented an important duration of MBT-A (1 year) compared to DBT-A
treatment component. No suicides were ob- (4 months). Within the limited observation
served, and overall there were few hospital period, we detected no significant group differ-
admissions or emergency department visits, ences with respect to the intensity of borderline

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 53 NUMBER 10 OCTOBER 2014 www.jaacap.org 1089
MEHLUM et al.

symptoms. Follow-up studies are needed to (such as self-harm and suicidal ideation) rather
clarify the long-term course of both this and other than an entire range of behaviors in the lives of
outcomes. multi-problem adolescents would be more
All treatments in this trial were delivered effective. &
at community child and adolescent psychiatric
outpatient clinics at no cost to the families, within
the framework of the universal health insurance
Clinical Guidance
system of Norway. Our patient sample was thus
probably less skewed with respect to socioeco-
 DBT-A is an effective treatment for self-harming ad-
nomic factors than studies in other contexts; this
olescents with traits of borderline personality disor-
strengthens the external validity of findings der in terms of reducing self-harm behavior, suicidal
toward similar urban treatment settings. Both ideation, and depressive symptoms.
DBT-A individual therapy sessions and the multi-
 DBT-A is feasible to administer at a high level of
family skills training were rated within the treatment fidelity in community child and adolescent
adherent range, showing that DBT-A can indeed psychiatric outpatient clinics.
be successfully delivered within a community
 DBT-A has a substantially shorter duration than most
mental health setting, not just in university comparable psychotherapeutic interventions
research clinics or by experts with many years of targeting self-harming adolescents.
DBT experience.
Although these findings are promising, this
study has several important limitations. The
study was adequately powered; however, the
sample size was small, and findings should be Accepted July 18, 2014.
interpreted with caution. The control condition, Drs. Mehlum, Ramberg, Haga, Grøholt and Mss. Tørmoen, Laberg,
and Diep are with the National Centre for Suicide Research and Pre-
EUC, was not a manualized treatment, and EUC vention, Institute of Clinical Medicine, University of Oslo, Norway.
therapists were not monitored for fidelity. Only Dr. Larsson is with the Norwegian University of Science and Technol-
DBT-A patients received skills-training group ogy, Trondheim, Norway. Dr. Stanley is with Columbia University,
New York City, and the University of Oslo, Norway. Dr. Miller is with
sessions; this implied a significant difference in Montefiore Medical Center and Albert Einstein College of Medicine,
the treatment intensity between interventions. New York City. Dr. Sund is with the Norwegian University of
Science and Technology and St Olav’s University Hospital, Trondheim,
Although our analyses gave few indications Norway.
that treatment intensity served as a mediator of The study was funded by grants from the Norwegian Directorate of
the association between treatment and out- Health, the South Eastern Regional Health Authority, the Extra-
Foundation for Health and Rehabilitation, and the University of Oslo.
comes, the possibility that this may have been
Lien M. Diep, MSc, served as the statistical expert for this research.
an important factor should not be neglected. As
The authors thank the patients, therapists, and leaders at the partici-
in most trials of self-harm, the patient sample pating clinics. They also thank Anne Brager-Larsen, MA, of the Oslo
was mostly female, and the sample was too University Hospital, and Fredrik Walby, MA, of the Diakonhjemmet
small to study gender differences in treatment Hospital, as dialectical behavior therapy (DBT) A team leaders; Sarah
Reynolds, PhD, of Columbia University, New York City, for providing
outcomes. DBT consultation team supervision; Ingrid Spurkland, MD, Inger H.
Among the study’s strengths are the absence of Vandvik, MD, PhD, Bjørn Koldsland, MD, of the University of Oslo,
and Svanhild Slaatto, Per O. Flaten Gunstad, Magnus Bjørkavoll-
dropout from the research, the application of Bergseth, Camilla Sivesind Mehlum, Ine Røed, Kristin Wickstrøm, and
rigorous procedures for data collection, the Trine Wiberg Dæhlie, undergraduates at University of Oslo, for con-
integrity of ratings and blinding, and the inde- ducting assessment interviews; Edvard Hauff MD, PhD, of the University
of Oslo, Johs Wiik, MD, of the Norwegian Institute of Public Health,
pendence of raters. The liberal inclusion criteria and John Eriksen, PhD, of Norwegian Social Research, for partici-
and the delivering of treatments in a community pating in the study steering group.
mental health setting with patients recruited from Disclosure: Drs. Mehlum, Ramberg, Haga, Larsson, Stanley, Miller,
Sund, Grøholt, and Mss. Tørmoen, Diep, and Laberg report no
a defined catchment area strengthen the external biomedical financial interests or potential conflicts of interest.
validity of the findings. Correspondence to Lars Mehlum, MD, PhD, National Centre for Sui-
Future studies, preferably with larger sam- cide Research and Prevention, Institute of Clinical Medicine, University
ples and long-term follow-up evaluations, are of Oslo, Sognsvannsveien 21, Bygg 12 N-0372, Oslo, Norway;
e-mail: [email protected]
needed to clarify whether these positive results
0890-8567/$36.00/ª2014 American Academy of Child and
will persist. Questions left to be addressed are Adolescent Psychiatry
what the active ingredients in DBT-A are, and http://dx.doi.org/10.1016/j.jaac.2014.07.003
whether focusing on specific symptom domains

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

1090 www.jaacap.org VOLUME 53 NUMBER 10 OCTOBER 2014
 EFFICACY OF DBT FOR SUICIDAL ADOLESCENTS

REFERENCES
1. Hawton K, Rodham K, Evans E, Weatherall R. Deliberate self 21. Miller AL, Rathus JH, Linehan MM. Dialectical Behavior Therapy
harm in adolescents: self report survey in schools in England. BMJ. with Suicidal Adolescents. New York: Guilford Press; 2007.
2002;325:1207-1211. 22. Woodberry KA, Popenoe EJ. Implementing dialectical behavior
2. Madge N, Hewitt A, Hawton K, et al. Deliberate self-harm within therapy with adolescents and their families in a community
an international community sample of young people: comparative outpatient clinic. Cogn Behav Pract. 2008;15:9.
findings from the Child and Adolescent Self-harm in Europe 23. James AC, Taylor A, Winmill L, Alfoadari K. A preliminary
(CASE) Study. J Child Psychol Psychiatry. 2008;49:667-677. community study of dialectical behavior therapy (DBT) with
3. Jacobson CM, Gould M. The epidemiology and phenomenology of adolescent females demonstrating persistent, deliberate self-harm
non-suicidal self-injurious behavior among adolescents: a critical (DSH). Child Adolesc Ment Health. 2008;13:7.
review of the literature. Arch Suicide Res. 2007;11:129-147. 24. Fleischhaker C, Bohme R, Sixt B, Bruck C, Schneider C, Schulz E.
4. Ystgaard M, Arensman E, Hawton K, et al. Deliberate self-harm Dialectical Behavioral Therapy for Adolescents (DBT-A): a clinical
in adolescents: comparison between those who receive help trial for patients with suicidal and self-injurious behavior and
following self-harm and those who do not. J Adolesc. 2009;32: borderline symptoms with a one-year follow-up. Child Adolesc
875-891. Psychiatry Ment Health. 2011;5:3.
5. Owens D, Horrocks J, House A. Fatal and non-fatal repetition of 25. Cooney E, Davis K, Thompson P, Wharewera-Mika J, Stewart J.
self-harm. Systematic review. Br J Psychiatry. 2002;181:193-199. Feasibility of Evaluating DBT for Self-harming Adolescents: a
6. Linehan MM. Cognitive-Behavioral Treatment of Borderline Per- Small Randomised Controlled Trial. Auckland, New Zealand: Te
sonality Disorder. New York: Guilford Press; 1993. Pou o Te Whakaaro Nui; 2010.
7. Soloff PH, Lynch KG, Kelly TM, Malone KM, Mann JJ. Charac- 26. Katz LY, Cox BJ, Gunasekara S, Miller AL. Feasibility of dialectical
teristics of suicide attempts of patients with major depressive behavior therapy for suicidal adolescent inpatients. J Am Acad
episode and borderline personality disorder: a comparative study. Child Adolesc Psychiatry. 2004;43:276-282.
Am J Psychiatry. 2000;157:601-608. 27. Rathus JH, Miller AL. Dialectical behavior therapy adapted for
8. Miller AL, Muehlenkamp JJ, Jacobson CM. Fact or fiction: diag- suicidal adolescents. Suicide Life Threat Behav. 2002;32:146-157.
nosing borderline personality disorder in adolescents. Clin Psy- 28. Pearson JL, Stanley B, King C, Fisher C. Issues to Consider in
chol Rev. 2008;28:969-981. Intervention Research with Persons at High Risk for Suicidality.
9. Winograd G, Cohen P, Chen H. Adolescent borderline symptoms Rockville, MD: National Institute of Mental Health; 2001.
in the community: prognosis for functioning over 20 years. J Child 29. Kaufman J, Birmaher B, Brent D, et al. Schedule for Affective
Psychol Psychiatry. 2008;49:933-941. Disorders and Schizophrenia for School-Age Children–Present
10. Wood A, Trainor G, Rothwell J, Moore A, Harrington R. Ran- and Lifetime Version (K-SADS-PL): initial reliability and validity
domized trial of group therapy for repeated deliberate self-harm data. J Am Acad Child Adolesc Psychiatry. 1997;36:980-988.
in adolescents. J Am Acad Child Adolesc Psychiatry. 2001;40: 30. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical
1246-1253. Interview for DSM-IV Personality Disorders, (SCID-II). Washing-
11. Hazell PL, Martin G, McGill K, et al. Group therapy for repeated ton, DC: American Psychiatric Press; 1997.
deliberate self-harm in adolescents: failure of replication of a 31. Achenbach TM. ASEBA. Manuals for TheYouth Self Report (11-18)
randomized trial. J Am Acad Child Adolesc Psychiatry. 2009;48: and the Child Behavior Checklist/6-18. Burlington, VT: Depart-
662-670. ment of Psychiatry. University of Vermont; 2001.
12. Green JM, Wood AJ, Kerfoot MJ, et al. Group therapy for adoles- 32. Shaffer D, Gould MS, Brasic J, et al. A Children’s Global Assess-
cents with repeated self harm: randomised controlled trial with ment Scale (CGAS). Arch Gen Psychiatry. 1983;40:1228-1231.
economic evaluation. BMJ. 2011;342:d682. 33. Linehan MM, Comtois KA. Lifetime Parasuicide Count. Seattle:
13. Rossouw TI, Fonagy P. Mentalization-based treatment for self- University of Washington; 1996.
harm in adolescents: a randomized controlled trial. J Am Acad 34. Beck AT, Schuyler D, Herman I. Development of suicide intent
Child Adolesc Psychiatry. 2012;51:1304-1313.e1303. scales. In: Beck AT, Resnick HLP, Lettiem D, Bowie MD, eds. The
14. Brent DA, McMakin DL, Kennard BD, Goldstein TR, Mayes TL, Prediction of Suicide. Philadelphia: Charles Press; 1974.
Douaihy AB. Protecting adolescents from self-harm: a critical re- 35. Reynolds WM, Mazza JJ. Assessment of suicidal ideation in inner-
view of intervention studies. J Am Acad Child Adolesc Psychiatry. city children and young adolescents: reliability and validity of the
2013;52:1260-1271. suicidal ideation questionnaire–JR. School Psych Rev. 1999;
15. Carter GL, Willcox CH, Lewin TJ, Conrad AM, Bendit N. Hunter 28:17-30.
DBT project: randomized controlled trial of dialectical behaviour 36. Angold A, Costello J, Messer SC. Development of a short ques-
therapy in women with borderline personality disorder. Aust NZ J tionnaire for use in epidemiological studies of depression in
Psychiatry. 2010;44:162-173. children and adolescents. Int J Methods Psychiatr Res. 1995;5:
16. Clarkin JF, Levy KN, Lenzenweger MF, Kernberg OF. Evaluating 237-249.
three treatments for borderline personality disorder: a multiwave 37. Montgomery SA, Asberg M. A new depression scale designed to
study. Am J Psychiatry. 2007;164:922-928. be sensitive to change. Br J Psychiatry. 1979;134:382-389.
17. Koons CR, Robins CJ, Tweed JL, et al. Efficacy of dialectical 38. Beck AT, Weissman A, Lester D, Trexler L. The measurement of
behavior therapy in women veterans with borderline personality pessimism: the Hopelessness Scale. J Consult Clin Psychol. 1974;
disorder. Behav Ther. 2001;32:371-390. 42:861-865.
18. Linehan MM, Armstrong HE, Suarez A, Allmon D, Heard HL. 39. Bohus M, Limberger MF, Frank U, Chapman AL, Kuhler T,
Cognitive-behavioral treatment of chronically parasuicidal Stieglitz RD. Psychometric properties of the Borderline Symptom
borderline patients. Arch Gen Psychiatry. 1991;48:1060-1064. List (BSL). Psychopathology. 2007;40:126-132.
19. Linehan MM, Comtois KA, Murray AM, et al. Two-year ran- 40. Pocock SJ. Clinical Trials. A Practical Approach. Chichester: John
domized controlled trial and follow-up of dialectical behavior Wiley and Sons; 1983.
therapy vs therapy by experts for suicidal behaviors and border- 41. STATA Statistical Software [computer program]. College Station,
line personality disorder. Arch Gen Psychiatry. 2006;63:757-766. TX: Stata Corporation; 2011.
20. Verheul R, Van Den Bosch LM, Koeter MW, De Ridder MA, 42. IBM SPSS Statistics for Windows [computer program]. Version 20.0.
Stijnen T, Van Den Brink W. Dialectical behaviour therapy for Armonk, NY: IBM Corporation; 2012.
women with borderline personality disorder: 12-month, rando- 43. Neacsiu AD, Rizvi SL, Linehan MM. Dialectical behavior therapy
mised clinical trial in the Netherlands. Br J Psychiatry. 2003;182: skills use as a mediator and outcome of treatment for borderline
135-140. personality disorder. Behav Res Ther. 2010;48:832-839.

JOURNAL OF THE AMERICAN ACADEMY OF C HILD & ADOLESCENT PSYCHIATRY

VOLUME 53 NUMBER 10 OCTOBER 2014 www.jaacap.org 1091