Acta Psychiatr Scand 2017: 135: 138–148 © 2016 John Wiley & Sons A/S.

6 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
All rights reserved ACTA PSYCHIATRICA SCANDINAVICA
DOI: 10.1111/acps.12664

A randomized trial of brief dialectical

behaviour therapy skills training in suicidal
patients suffering from borderline disorder
McMain SF, Guimond T, Barnhart R, Habinski L, Streiner DL. A Shelley F. McMain1,2,
randomized trial of brief dialectical behaviour therapy skills training in Tim Guimond1,2,
suicidal patients suffering from borderline disorder. Ryan Barnhart1,3, Liat Habinski4,
David L. Streiner2,5
Objective: Evidence-based therapies for borderline personality disorder 1
Centre for Addiction and Mental Health, Toronto, ON,
(BPD) are lengthy and scarce. Data on brief interventions are limited, 2
Department of Psychiatry, University of Toronto,
and their role in the treatment of BPD is unclear. Our aim was therefore Toronto, ON, 3York University, Toronto, ON, 4Royal
to evaluate the clinical effectiveness of brief dialectical behaviour Alexandra Hospital, Edmonton, AB and 5Department of
therapy (DBT) skills training as an adjunctive intervention for high Psychiatry and Behavioural Neurosciences, McMaster
suicide risk in patients with BPD. University, Hamilton, ON, Canada
Method: Eighty-four out-patients were randomized to 20 weeks of
DBT skills (n = 42) or a waitlist (WL; n = 42). The primary outcome
was frequency of suicidal or non-suicidal self-injurious (NSSI) episodes.
Assessments were conducted at baseline 10, 20 and 32 weeks.
Results: DBT participants showed greater reductions than the WL Key words: randomized controlled trial; personality
disorder; suicide; self-harm; psychotherapy
participants on suicidal and NSSI behaviours between baseline and
32 weeks (P < 0.0001). DBT participants showed greater improvements Shelley McMain, Centre for Addiction and Mental
than controls on measures of anger, distress tolerance and emotion Health, 33 Russell Street, Toronto, ON, M5S 2S1,
Canada. E-mail: [email protected]
regulation at 32 weeks.
Conclusions: This abbreviated intervention is a viable option that may
be a useful adjunctive intervention for the treatment of high-risk
behaviour associated with the acute phase of BPD. Accepted for publication October 14, 2016

Significant outcomes
• Support for the effectiveness of brief skills-based group interventions for borderline personality disor-
der.
• Superior improvements in the reduction of self-destructive (e.g. suicidal and self-harm) behaviours,
aggressive behaviour (e.g. anger) and coping skills (e.g. distress tolerance and emotion regulation)
amongst those in the dialectical behaviour therapy skills training group, which were maintained at
follow-up.

Limitations
• Ancillary treatments were not excluded.
• Use of self-report for primary outcome measures.

intent), which is a strong risk factor for suicide, is

Introduction
reported by 69–80% (4, 5). Apart from the per-
Borderline personality disorder (BPD) is a preva- sonal and social impacts, self-harm contributes to
lent disorder that has an estimated lifetime preva- a significant economic burden due to loss of pro-
lence of 6% (1, 2). The suicide rate amongst ductivity and high healthcare utilization.
individuals with BPD is estimated to be as high as Dialectical behaviour therapy (DBT) has been
10% (3), while self-harm (with or without suicide used successfully in treating BPD, with several

138
 Dialectical behaviour therapy skills training

well-controlled trials providing robust evidence for intervention for high-risk suicidal individuals with
its effectiveness (6). The majority of research sup- borderline personality disorder in a well-controlled
porting the efficacy of DBT for the treatment of and well-powered pragmatic trial, with the primary
BPD pertains to the standard model, which is a 12- outcome being the difference between the dialecti-
month, multi-modal treatment, including individ- cal behaviour therapy group and an active waitlist
ual sessions, skills group, phone coaching and a control group in the frequency of suicidal and non-
therapist consultation team. Access to standard suicidal self-injury behaviours. At this stage,
DBT is restricted due to limited resources and a because there are few if any similar trials, a first-
shortage of well-trained clinicians, resulting in generation study that includes a waitlist control
lengthy waitlists. This situation is not unique; condition as a comparator is needed to address the
other BPD-specific treatments are similarly question of whether dialectical behaviour therapy
restricted. Because standard DBT is perceived as skills training itself is effective. Many borderline
costly and complex to deliver to all patients who personality disorder specialist programmes in the
need it, many clinical settings are delivering only community that offer dialectical behaviour therapy
the skills training component, even though this skills training as a stand-alone intervention have
practice lacks adequate empirical support. lengthy waitlists and lack substitute treatments. In
DBT skills-only treatment for BPD has been effect, we tested dialectical behaviour therapy skills
evaluated in two randomized controlled studies. training against a waitlist, because it is currently
Soler et al. (7) found 13 weeks of DBT skills train- the only available option in many real-world clini-
ing to be superior to standard group therapy in cal settings. Significant effects supporting the supe-
improving BPD symptoms, depression, anxiety, riority of dialectical behaviour therapy skills
anger and affect instability. In the other RCT, training compared to a waitlist control would be
which specifically recruited suicidal and self-injur- evidence that improvements in outcome are due to
ing patients with BPD, Linehan et al. (8) assigned treatment. In the light of the need for effective
patients to a one-year intervention of either stan- interventions that are easily transportable and
dard DBT, the skills component of DBT plus cost-effective, information about the effectiveness
intensive case management, or individual DBT ses- of a brief dialectical behaviour therapy skills train-
sions only. The results showed that the DBT ing intervention is highly relevant.
modes with skills training (e.g. standard DBT or
DBT skills training plus intensive care manage-
Material and methods
ment) were more effective than individual therapy
in reducing suicidal and NSSI behaviours. Addi- This two-arm, single-blinded, prospective, ran-
tional evidence to support the effectiveness of DBT domized controlled trial was designed to evaluate
skills-only treatment is needed. the clinical effectiveness of a 20-week DBT skills
Several trials have evaluated the outcome of training group compared to an active waitlist
brief skills-only-based interventions for BPD (WL) group in which ancillary treatments were
patients and have found moderate to large effect unrestricted for both groups. The study was con-
sizes (9–12). However, it is unclear whether these ducted between October 2010 and March 2012 at
abbreviated skills-only interventions are effective the Centre for Addiction and Mental Health
in high-risk individuals, because most of the trials (CAMH), a teaching hospital affiliated with the
did not have suicide or non-suicidal self-injury University of Toronto. The study was approved by
(NSSI) as a primary outcome or focus. For exam- the CAMH Research Ethics Board, and written
ple, high-risk patients have been viewed as requir- informed consent was obtained from participants
ing comprehensive treatment that includes prior to enrolment. The costs of treatment were
individual therapy and between-session phone covered by the Canadian public healthcare system.
coaching to adequately manage suicide risk (13). Study inclusion criteria were as follows: (i) meet-
Brief DBT skills training may be a viable treatment ing the criteria for BPD as defined in the Diagnos-
alternative for suicidal individuals with BPD. It tic and Statistical Manual Version IV (DSM-IV)
would be of enormous practical and theoretical (14), (ii) 18–60 years of age, (iii) two suicidal and/
value to answer this question. or NSSI episodes in the past 5 years, with one
occurring within 10 weeks prior to enrolment and
(iv) able to understand written and spoken
Aims of the study
English. To maximize external validity, exclusion
The aim of this study was to assess the effectiveness criteria were limited to the following: (i) meeting
of a brief course of dialectical behaviour therapy DSM-IV criteria for a psychotic disorder, bipolar I
skills training designed as an adjunctive disorder or dementia, (ii) evidence of an organic

139
McMain et al.

brain syndrome or mental retardation based on ‘adherent’. Adherence ratings were conducted by
clinical interview and (iii) participation in a DBT one well-trained and reliable coder on a random
programme within the past year. sample of 10% of all videotaped group sessions.
The Structured Clinical Interview I for the Participants assigned to the waitlist control con-
DSM-IV (SCID-I) (15) and the International Per- dition remained on the list for five months (e.g.
sonality Disorder Exam (IPDE) (16) were used to end of follow-up). At the end of the study, they
assess Axis I and Axis II diagnoses respectively. were offered a place in treatment. During this wait
High inter-rater reliability was observed for the period, participants could continue with treat-
number of BPD symptoms (intra-class correlation ment-as-usual care (medication management or
coefficient = 0.97). Participants were assessed by other psychosocial treatments).
two doctoral-level psychology students and one
master’s-level clinician who were well trained on
Outcomes
the study instruments and were blinded to treat-
ment assignment, while treatment history inter- Study assessments were conducted at baseline
views were conducted by two research assistants (pre-intervention), 10 weeks, 20 weeks (end of
who were not blinded to treatment assignment. intervention) and 32 weeks (3-month follow-up).
Following completion of baseline assessments, Participants were compensated $10 per hour for
participants were assigned to groups using a stan- completing the assessments.
dard random block design in block sizes of four. The primary outcome, frequency of suicidal
The statistician prepared 42 envelopes, each con- and/or NSSI episodes, was assessed using two
taining two allocations to each of the conditions in instruments: the clinician-administered Lifetime
random order. Suicide Attempt Self-Injury Interview (LSASI; for-
merly Lifetime Parasuicide Count, M. M. Linehan,
K. A. Comtois, unpublished data, 1996) and the
Treatment and therapists
self-report Deliberate Self-Harm Inventory
The DBT group skills training consisted of the (DSHI) (20). The LSASI has similar items to the
manualized approach developed by Linehan Suicide Attempt Self-Injury Interview (SASII),
(17, 18), adapted to a 20-week curriculum in which which has good inter-rater reliability (0.87–0.98).
groups meet for 2 h weekly. A description of the It has been used in DBT trials to assess the topog-
20-week skills curriculum can be found in Line- raphy of suicidal and NSSI behaviours (8, 21–24).
han’s skills manual (19). The training uses a psy- The DSHI is a 17-item self-report measure that
cho-educational focus to enhance capabilities. The assesses the method, frequency and medical sever-
following five modules were covered: mindfulness, ity of deliberate self-harm without suicidal intent.
emotion regulation, distress tolerance, interper- It has high internal consistency (alpha = 0.82),
sonal effectiveness and dialectics. Prior to the first adequate test–retest reliability and good construct
group meeting, participants attended a 90-min validity. The frequency of suicide attempts and
individual orientation session. Skills group leaders NSSI episodes was computed from participants’
were not available to provide crisis coaching out- responses to the LSASI and the DSHI instruments
side of skills group sessions. Participants were respectively.
encouraged to have a therapist or another individ- Secondary outcomes were changes in healthcare
ual (e.g. family practitioner, spiritual counsellor, utilization, BPD symptoms and coping. Healthcare
family member) who could provide crisis support. utilization was assessed using a semistructured
Additionally, participants were offered a list of interview, the Treatment History Interview-2
resources for crisis support (e.g. crisis call lines, (THI-2) (25), to determine the number of emer-
distress centres). To increase external validity, gency department (ED) visits, psychiatric hospital
there were no restrictions on ancillary treatments. admissions and use of medications. Symptoms
Treatment was delivered by five therapists were assessed using the Borderline Symptom List-
(PhD = 2; MSW = 3) with an average of 8.4 years 23 (BSL-23) (26), a self-report scale to assess bor-
(SD = 4.66) of experience in facilitating DBT derline typical symptomatology; the State-Trait
groups for patients with BPD. Therapists attended Anger Expression Inventory (STAXI) (27); the
a weekly consultation team. Symptom Checklist-90-Revised (SCL-90-R) (28)
Treatment fidelity was evaluated using the DBT to measure general psychiatric symptoms; the Bar-
Global Rating Scale (GRS; M. M. Linehan, rett Impulsiveness Scale-11 (BIS-11) (29); the Beck
unpublished data, 1993), a 66-item instrument that Depression Inventory-II (BDI-II) (30); and the
codes adherence to DBT on a five-point scale, with Social Adjustment Scale–Self-Report (SAS-SR)
an overall score of 4 or higher indicating (31). Coping was assessed using the Difficulties in

140
 Dialectical behaviour therapy skills training

Emotion Regulation Scale (DERS) (32), the Dis- This was also evident in the analysis of model
tress Tolerance Scale (DTS) (33) and the Kentucky residuals. To account for this, each linear model
Inventory of Mindfulness Scale (KIMS) (34). was estimated under the heteroskedastic assump-
Participants who dropped out prematurely were tion, with means and variances estimated freely for
requested to complete the Reasons for Early Ter- each group (36).
mination from Treatment Questionnaire (35) to Each model was reanalysed using preselected
indicate why they had discontinued. The drop-out covariates that are known or theorized to affect the
definition was modified from what is used in a outcome, and the results were compared with the
standard one-year DBT programme (e.g. four con- primary models outlined above to determine
secutively missed individual or group sessions) whether including covariates caused meaningful
because this was a shorter and single-mode inter- changes in the overall findings. Conditional esti-
vention. Treatment drop-out was defined as failure mates of the means and variances were not found
to participate in three consecutive scheduled group to deviate in the presence of the covariates, and so
sessions or five group sessions in total. the unconditional models were selected to not sup-
press any possible estimates of the treatment effect.
Using these methods, the statistics are based on
Power and statistical analysis of change
available case analyses where the full information
The power analysis was conducted using the rate of the data set can be employed, as the models can
of change in frequency of self-harm episodes, accommodate unbalanced and missing data. All
based on data from the only available relevant models were estimated using restricted maximum
study evaluating a time-limited skills training likelihood (REML), which is preferable to full
group treatment for self-harm and BPD (9). We maximum likelihood when the samples are small.
expected to see a group difference of 20% in the All comparisons of conditional means were
frequency of suicidal and/or NSSI episodes. Set- performed with Wald v2 tests, and all reported
ting the alpha level at 5% and the beta level at P-values were adjusted for family-wise type I error
0.20, and estimating a drop-out rate of 30%, it was inflation using the Holm’s sequential Bonferroni
determined that a sample of 84 participants adjustment.
(DBT = 42; WL = 42) was required to show statis-
tically significant differences.
Results
All analyses were conducted on the intent-to-
treat sample (N = 84). To assess the effectiveness A total of 140 prospective subjects were screened,
of our randomization process, between-group and 84 eligible participants were randomly
comparisons of baseline characteristics on all mea- assigned to DBT skills training (n = 42) or a WL
sures were conducted, using t-tests for continuous control (n = 42). The subject flow is shown in
variables and chi-square tests for categorical/nomi- Fig. 1.
nal variables. Baseline characteristics are shown in Table 1.
Several of the primary count measures, such as After correcting for multiple testing, there were no
suicidal and NSSI behaviours, hospitalization between-group differences in demographic charac-
days, and ED visits, were non-normally distributed teristics, clinical characteristics or number of suici-
and were therefore analysed using multilevel longi- dal or NSSI episodes.
tudinal generalized linear models (MLGLM). For
suicidal and NSSI variables, a Poisson distribution
Attrition and treatment adherence
was used. To control for heterogeneous and incon-
sistent responses within participants, both random Of the 42 participants assigned to DBT skills train-
intercepts and occasion-level random effects were ing, 29 (71%) completed the treatment and 13
estimated for and within each participant (36, 37). (31%) dropped out prematurely. Treatment com-
For hospitalization days and ED visits, due to pleters attended a mean of 17.9 sessions
excessive zero inflation, the data were collapsed to (SD = 1.6), while those who dropped out attended
a binary outcome (hospitalizations yes/no and ED a mean of 5.62 (SD = 5.9). The most commonly
visits yes/no) and analysed assuming a logistic dis- reported reasons for dropping out were ‘didn’t
tribution. think sessions were helpful’ (n = 4) and ‘time
Multilevel linear growth curve models were used problems’, ‘transportation problems’ and ‘medical
to analyse normally distributed secondary outcome reasons’ (n = 3 each).
measures, with observations nested within partici- In terms of missing data, study follow-up assess-
pant (36). Estimates of variance over time between ments were completed by 74 participants (88.1%).
treatment groups indicated heteroskedasticity. Participants completed a mean of 2.61 of the three

141
McMain et al.

Enrolment

140 assessed for eligibility

56 excluded
39 did not meet inclusion criteria
6 declined to parcipate
11 other reasons

84 randomized

Allocaon
42 allocated to 20 -week DBT group
42 allocated to waitlist
30 completed Intervenon

Follow-up
5 lost to follow-up 3 lost to follow-up

Fig. 1. CONSORT diagram of

participants randomized to dialectical
Analysis behaviour therapy skills training or a
waitlist control for borderline
42 included in the primary analysis 42 included in the primary analysis personality disorder. [Colour figure can
be viewed at wileyonlinelibrary.com]

postbaseline study assessments (DBT group, differences at any time point (all v2(1) ≤ 1.179,
2.48+/ 0.16; waitlist, 2.74+/ 0.12). There was no P > 0.28) (See Table 2 for psychosocial treatments
evidence that missing data patterns were biased by utilized by participants over time).
group membership (Fisher’s exact test, P = 0.26). Amongst the specific forms of psychosocial
At baseline, a total of 71 patients (86%) were treatments patients were receiving over time, two
taking psychotropic medications, with a mean of differences were observed. After adjusting for
1.79  1.41 medications per participant. There family-wise error rates, there were no statistically
were no significant between-group differences in significant differences between groups in the
either the number of patients on medication proportions receiving treatment, except for group
(DBT = 33/42; WL = 38/42; v2(1) = 1.67, psychotherapy at 10 weeks (DBT = 26/36;
P = 0.20) or the mean number of medications WL = 10/35; v2(1) = 13.53, P < 0.001) and
taken (DBT = 1.52  0.20; WL = 2.05  0.22; 20 weeks (DBT = 24/32; WL = 7/38; v2(1) ≤
t(80) = 1.73, P > 0.05). At 20 weeks, a total of 57 22.54, P < 0.001). However, this result was
patients (81%) were taking medication and were expected because it included the DBT skills treat-
averaging 1.62  1.67 medications, with the DBT ment.
group reporting both fewer patients on medication Treatment adherence ratings were conducted on
(23/32) compared to the WL patients (34/38) and 10% (n = 22) of sessions. The mean score of 4.44
fewer medications (1.52  0.20 vs. 2.05  0.22, (SD = 0.11) fell within the ‘adherent’ range.
t(80) = 2.10, P = 0.04). There was no significant
group difference in the average number of medica-
Outcome analyses
tions at the 32-week follow-up (t(80) = 0.53,
P = 0.60). Table 3 shows the results of all outcome analyses,
At baseline, a total of 71 patients (85%) and Fig. 2 illustrates the major findings.
reported that they were receiving some form of
psychosocial treatment from a therapist
Suicidal and/or NSSI behaviours
(DBT = 35/40; WL = 36/39). At 10, 20 and
32 weeks, these numbers were 76% (32/36), 64% There were no completed suicides in either group.
(27/32) and 60% (25/37) for the DBT group, and Based on MLGLM analyses, the DBT group
67% (28/35), 67% (28/38) and 57% (24/39) for the showed statistically greater reductions in the fre-
WL participants. There were no between-group quency of suicidal and self-harm episodes as

142
 Dialectical behaviour therapy skills training

Table 1. Baseline demographic and diagnostic characteristics for 84 out-patients with borderline personality disorder

Entire sample
DBT skills Waitlist
(n = 42) (n = 42) (n = 84) P value

Women 35 (83.3) 31 (73.8) 66 (78.6) P = 0.29§

Marital status
Married, common law 3 (7.1) 8 (19.1) 11 (13.1)
P = 0.26§
Separated, divorced, widowed 6 (14.3) 6 (14.3) 12 (14.29)
Never married 33 (78.6) 28 (66.7) 61 (72.6)
Education
<High school 4 (9.52) 1 (2.4) 5 (5.0)
High school graduate 5 (11.9) 3 (7.1) 8 (9.5) P = 0.44§
Some college or technical school 13 (31.0) 16 (38.1) 29 (34.5)
College/university graduate 20 (47.6) 22 (52.4) 42 (50.0)
Employment
Unemployed 10 (23.8) 10 (23.8) 20 (23.8)
Full time 9 (21.4) 14 (33.3) 23 (27.4)
Part time 2 (4.8) 4 (9.5) 6 (7.1) P = 0.51§
Full-time student 8 (19.1) 3 (7.1) 11 (13.1)
Disabled 5 (11.9) 5 (11.9) 10 (11.9)
Receiving public assistance 8 (19.1) 6 (14.3) 14 (16.7)
Annual income
<$15 000 18 (42.9) 15 (35.7) 33 (39.3)
Between $15 000 and $29 000 4 (9.5) 4 (9.5) 8 (9.5) P = 0.61§
Between $30 000 and $49 000 3 (7.14) 8 (19.1) 11 (13.1)
>$50 000 8 (19.1) 7 (16.7) 15 (17.9)
No answer/refused to answer 9 (21.4) 8 (19.1) 17 (20.2)
Lifetime DSM-IV axis I diagnoses
Major depressive disorder 18 (42.9) 17 (40.4) 35 (41.7) P = 0.83§
Panic disorder 2 (4.8) 3 (7.1) 5 (6.0) P = 0.65§
Post-traumatic stress disorder 5 (11.9) 6 (14.3) 11 (13.1) P = 0.75§
Any anxiety disorder 23 (54.8) 24 (57.1) 47 (56.0) P = 0.83§
Substance abuse 4 (9.5) 2 (4.8) 6 (7.1) P = 0.40§
Substance dependence 11 (26.2) 13 (31.0) 24 (28.6) P = 0.63§
Any eating disorder 2 (4.8) 3 (7.14) 5 (6.0) P = 0.65§
Current DSM-IV axis I and axis II diagnoses†
Major depressive disorder 23 (54.8) 20 (47.6) 43 (51.2) P = 0.51§
Panic disorder 4 (9.5) 8 (19.1) 12 (14.3) P = 0.21§
Post-traumatic stress disorder 8 (19.1) 11 (26.2) 19 (22.6) P = 0.43§
Any anxiety disorder 21 (50.0) 30 (71.4) 51 (60.7) P = 0.04*,§,¶
Substance abuse 7 (16.7) 4 (9.5) 11 (13.1) P = 0.33§
Substance dependence 24 (57.1) 23 (54.8) 47 (56.0) P = 0.83§
Any eating disorder 5 (11.9) 8 (19.1) 13 (15.5) P = 0.37§
Mean (SD) Mean (SD) Mean (SD)
Current axis I disorders 2.14 (0.29) 2.36 (0.29) 2.25 (0.20) P = 0.60‡
Lifetime axis I disorders 3.0 (0.33) 2.62 (0.25) 2.81 (0.21) P = 0.36‡
Lifetime suicide attempts 7.11 (14.5) 115 (694) 61.1 (491) P = 0.32‡
Age, mean (SD) 27.29 (7.45) 32.05 (9.06) 29.67 (8.62) P = 0.01‡

DBT, dialectical behaviour therapy; DSM-IV, Diagnostic and Statistical Manual, Version 4.
*Indicates significant difference (P < 0.05), Χ2(1) = 5.26.
†Five participants were missing data, proportions are conservative.
‡Based on independent t-test.
§Based on chi-square.
¶After correcting for multiple testing, the difference between groups on anxiety disorders is not significant.

measured by the LSASI at 32 weeks (P < 0.04). favoured the DBT group at 10 weeks
On the DSHI, the between-group differences of the (P < 0.001) and 20 weeks (P < 0.001); how-
frequency of NSSI were in the same direction and ever, these differences were not apparent at
approached, but failed to meet significance at 32 weeks (P > 0.48). While there were margin-
32 weeks (P = 0.08). ally significant between-group differences in the
number of ED visits at 20 weeks (P > 0.06)
favouring the DBT group, there was no evi-
Healthcare utilization
dence that the differences between groups were
Based on MLGLM analyses, between-group dif- statistically significant at 20 weeks or at
ferences of the number of hospital admissions 32 weeks.

143
McMain et al.

Table 2. Psychosocial treatments utilized by 84 out-patients with borderline personality disorder (By Group Assignment)

Baseline 10 weeks 20 weeks 32 weeks

WL DBT WL DBT WL DBT WL DBT

Type of psychosocial treatment n (%) n (%) n (%) n (%)

Individual psychotherapy or counselling 36 (92%) 33 (83%) 26 (74%) 24 (67%) 27 (71%) 24 (75%) 20 (51%) 24 (65%)
Group psychotherapy or counselling* 18 (46%) 20 (50%) 10 (29%) 26 (72%) 7 (18%) 24 (75%) 13 (33%) 9 (24%)
Couples, marital or family psychotherapy or counselling 10 (26%) 6 (15%) 3 (9%) 1 (3%) 3 (8%) 2 (6%) 3 (8%) 1 (3%)
Case management 10 (26%) 8 (20%) 6 (17%) 2 (6%) 11 (29%) 2 (6%) 6 (15%) 2 (5%)
Day treatment 8 (21%) 11 (28%) 5 (14%) 5 (14%) 4 (11%) 1 (3%) 4 (10%) 1 (3%)
Job skills or vocational counselling 2 (5%) 1 (3%) 5 (14%) 0 (0%) 2 (5%) 0 (0%) 0 (0%) 2 (5%)
12 step group, spiritual counselling or direction 6 (15%) 6 (15%) 5 (14%) 2 (6%) 6 (16%) 4 (13%) 6 (15%) 5 (14%)

All observed percentages engaged are calculated from (n/N) where n is the number reported engaged and N is the available cases for that group at the specific time.
*Count includes experimental treatment (e.g. DBT skills training group).

both statistically reliable and clinically significant,

Coping skills
were 47.1% and 20.6%, respectively, while those in
Mixed-effects linear growth curve analyses of the the WL group were 41.0% and 20.5% respectively.
measures of coping skills (distress tolerance, emo- Analyses of group differences revealed significant
tion regulation) revealed significantly greater differences in the odds of achieving both statisti-
improvements in the DBT group compared to the cally reliable change and clinically significant
WL group on distress tolerance and emotion regula- change at 20 weeks (OR = 3.44, z = 2.25,
tion, at all time points. On mindfulness, there were P = 0.024), but not at 32 weeks. This result was
no between-group differences at any time point. also confirmed on the differences in the odds of
achieving statistically reliable change only at
20 weeks (OR = 3.21, z = 2.33, P = 0.020), but
Mental health outcomes
again not at 32 weeks (See Fig. S3).
Mixed-effects linear growth curve analyses
revealed significantly lower levels of anger in the
Discussion
DBT group than the WL group at all time
points. The DBT group had lower levels of anger The vast majority of individuals with BPD are
at baseline; however, the DBT group made unable to access specialist treatment, and there is an
greater reductions in anger over time, while the urgent need for less resource-intensive options. The
WL group did not (slope difference = 0.25, aim of this RCT was to evaluate the effectiveness of
z = 2.50, P = 0.013). The DBT group showed a brief DBT skills training group programme
significantly greater gains on social adjustment, designed as an adjunctive intervention for the treat-
symptom distress and borderline symptoms at ment of high-risk suicidal individuals with BPD. To
20 weeks; however, these group differences were our knowledge, this is the first pragmatic, randomly
not maintained at 32 weeks. There were no sig- controlled trial of an abbreviated format of DBT
nificant group differences on impulsivity at any skills training in this patient population. The results
time points. indicate that brief interventions do not need to be
restricted to individuals with mild symptoms and
may be beneficial to those at high risk for suicide.
Clinically significant change
The DBT group showed superior improvements
Jacobson et al.’s (38) two-fold criteria were applied in the reduction of self-destructive (e.g. suicidal and
to assess clinically significant improvement, as self-harm) behaviours. DBT participants also exhib-
assessed on the SCL-90-R. In the DBT group, ited significantly greater gains in aggressive beha-
56.3% of all participants showed changes from viour (e.g. anger) and in learning coping skills (e.g.
baseline to 20 weeks that were statistically reliable, distress tolerance and emotion regulation), com-
and 43.8% fulfilled the criteria for change that was pared to the control group. Improvements on these
both statistically reliable and clinically significant. outcomes were durable postdischarge, which sug-
In the WL group, these percentages were 28.9% gests that DBT skills training was therapeutic.
and 18.4% respectively. From baseline to Although DBT participants showed superior out-
32 weeks, the percentages in the DBT group who comes on healthcare utilization (e.g. hospital days
showed changes that were statistically reliable, or and ED visits), social adjustment, symptom distress

144
 Dialectical behaviour therapy skills training

Table 3. Outcomes for 84 individuals with borderline personality disorder, by Group Assignment

Treatment assignment

DBT WL
Outcome measures (n = 42) (n = 42) Between-group difference

Count/dichotomous outcomes (Incident rate/odds ratio) Wald v 2

P

# of Suicidal and self-injurious episodes, mean (SD)** (DSHI)

Baseline 9.68 (25.89) 10.12 (29.73)
10 weeks 3.32 (7.97) 5.12 (12.91)
20 weeks 1.14 (3.26) 2.59 (6.90) 4.77 <0.09
32 weeks 0.32 (1.27) 1.14 (3.94) 5.32 <0.08
# of suicidal and self-injurious episodes, mean (SD)** (LSASI)
Baseline 9.06 (8.31) 8.33 (7.62)
10 weeks 5.07 (4.26) 5.76 (4.80)
20 weeks 2.84 (2.36) 3.96 (3.47) 2.73 < 0.30
32 weeks 1.41 (1.35) 2.56 (2.40) 6.71 < 0.04
Proportion of participants with emergency room visits, mean (SD)
Baseline 0.61 (0.36) 0.32 (0.34)
10 weeks 0.29 (0.32) 0.19 (0.26)
20 weeks 0.11 (0.20) 0.25 (0.31) 5.33 <0.06
32 weeks 0.25 (0.30) 0.16 (0.25) 1.99 <0.16
Proportion of participants hospitalized, mean (SD)
Baseline 0.29 (0.31) 0.18 (0.21)
10 weeks 0.04 (0.10) 0.0.18 (0.20)
20 weeks 0.02 (0.07) 0.14 (0.18) 13.9 <0.001
32 weeks 0.14 (0.20) 0.11 (0.16) 0.5 <0.48

Normally distributed outcomes Wald v2 P Cohen’s d

Borderline symptoms checklist (BSL), mean (SD)

Baseline 56.35 (16.51) 58.75 (19.64) 0.32
10 weeks 45.03 (13.74) 53.61 (17.70)
20 weeks 33.72 (18.70) 48.48 (22.21) 8.98 <0.01
32 weeks 41.08 (22.41) 45.99 (26.27) 0.75 <0.77
Anger (STAXI, Anger Expression Out Scale Score), Mean (SD)
Baseline 38.73 (9.88) 45.22 (9.53) 0.8
10 weeks 34.23 (8.50) 43.20 (8.71)
20 weeks 29.73 (9.27) 41.18 (10.68) 23 <0.001
32 weeks 30.29 (10.96) 40.43 (12.16) 14.1 <0.001
Symptom distress (SCL-90R, Total Score), mean (SD)
Baseline 1.96 (0.59) 2.10 (0.68) 0.41
10 weeks 1.63 (0.54) 1.97 (0.64)
20 weeks 1.30 (0.65) 1.84 (0.76) 10.3 <0.005
32 weeks 1.47 (0.76) 1.69 (0.87) 1.33 <0.50
Barratt Impulsiveness Scale-11 (BIS-11), Mean (SD)
Baseline 57.82 (9.04) 55.83 (9.98) 0.08
10 weeks 55.31 (7.50) 56.02 (7.52)
20 weeks 52.79 (9.72) 56.20 (8.54) 2.29 < 0.52
32 weeks 53.32 (11.35) 55.16 (9.10) 0.57 < 0.90
Social adjustment scale–self-report (SAS-SR), mean (SD)
Baseline 2.84 (0.46) 2.84 (0.53) 0.45
10 weeks 2.67 (0.43) 2.92 (0.50)
20 weeks 2.50 (0.56) 2.88 (0.59) 7.49 <0.02
32 weeks 2.60 (0.70) 2.87 (0.65) 2.82 <0.19
Depression (BDI), mean (SD)
Baseline 32.68 (10.95) 36.70 (11.46) 0.32
10 weeks 27.72 (9.59) 33.21 (11.02)
20 weeks 22.76 (12.55) 29.73 (13.50) 4.84 < 0.08
32 weeks 27.94 (16.08) 29.50 (15.71) 0.24 < 0.62
Difficulties in emotion regulation scale (DERS), mean (SD)
Baseline 131.43 (17.84) 132.80 (16.79) 0.5
10 weeks 118.99 (15.55) 129.75 (15.76)
20 weeks 106.55 (20.22) 126.70 (18.76) 18 <0.001
32 weeks 110.63 (26.87) 128.06 (20.89) 9.11 <0.01
Distress Tolerance Scale (DTS), Mean (SD)
Baseline 5.11 (2.49) 4.45 (2.34) 0.56

145
McMain et al.

Table 3. (Continued)

Normally distributed outcomes Wald v2 P Cohen’s d

10 weeks 6.44 (2.16) 4.95 (1.87)

20 weeks 7.77 (2.93) 5.45 (2.49) 12.5 <0.005
32 weeks 7.81 (3.64) 5.28 (2.75) 11.1 <0.005
Kentucky Inventory of Mindfulness Skills (KIMS), Mean (SD)
Baseline 101.46 (19.21) 105.42 (15.51) 0.19
10 weeks 109.14 (15.57) 106.48 (14.87)
20 weeks 116.81 (17.53) 107.53 (17.41) 4.65 <0.2
32 weeks 114.6 (21.08) 107.76 (20.25) 1.91 <0.6

**Means reported are the fixed-effect marginal mean incident rates after adjusting for overdispersion between and within participants. DHSI, Deliberate Self-harm Inventory;
LSASI, Lifetime Suicide and Self-Injury Interview; BSL, Borderline Symptom Checklist; STAXI, State-Trait Anger Expression Inventory; SCL-90-R, Symptom Checklist 90-Revised;
BIS-11, Barratt Impulsiveness Scale; SAS-SR, Social Adjustment Scale–Self Report; BDI-II, Beck Depression Inventory-II; DERS, Difficulties in Emotion Regulation Scale; DTS,
Distress Tolerance Scale; KIMS, Kentucky Inventory of Mindfulness Skills.

Fig. 2. Outcomes for dialectical

behaviour therapy (N = 42) and
waitlist (N = 42) groups over
32 weeks after randomization.
Estimated marginal means and 95%
confidence intervals over specific
outcomes. DSHI, Deliberate Self-Harm
Inventory; LSASI, Lifetime Suicide
Attempt Self-Injury. Estimates of the
marginal means were derived from
multilevel generalized linear models.
Estimates are incident rate ratios for
the self-harm measures and
probabilities for the occurrence of
Emergency Room Visits and
Hospitalizations.

and borderline symptoms at the end of treatment, at This study provides additional support for the
32 weeks the gains on these outcomes either dimin- effectiveness of abbreviated interventions for BPD.
ished or the waitlist participants had caught up. Our findings are consistent with those of other
The study findings lend support to the theory studies that demonstrate the effectiveness of brief
that the development of coping skills may be a skills-based group interventions (10, 11, 40).
critical ingredient in DBT that accounts for Research on interventions for individuals at high
improvements on outcomes, an idea that is sup- risk for suicide is lacking, and evidence of strate-
ported by one DBT study that found that coping gies that may be successful for patients with severe
skills mediate outcomes (39). Our findings high- symptoms is needed.
light the benefits of DBT skills training and are Consistent with other studies in which brief
in line with the results of a recent component skills training interventions of BPD are best
analysis study by Linehan et al. (8), demonstrat- described as an adjunct to other treatment,
ing the superiority of DBT with skills training vs. receipt of other treatments was unrestricted, and
DBT without skills training for reducing self- the majority of participants received some other
harm behaviours, amongst individuals with at form of ancillary treatments during the 32 weeks.
least one episode of self-harm during the year of It is not known how these other treatments
treatment. Our findings extend knowledge about impacted the effects of the skills training, and we
the effectiveness of DBT skills training by cannot rule out the possibility that the effects
demonstrating that an abbreviated format of that were observed are related to these con-
DBT is useful for reducing symptoms associated founds. However, utilization of other treatments
with the acute phase of BPD. did not differ between the groups, and

146
 Dialectical behaviour therapy skills training

participants in the WL condition showed fewer unknown. Finally, the follow-up was for only
improvements than did those in the DBT group, three months, which may not have been long
suggesting that the most plausible explanation is enough to address questions about the durability
that the observed effects are attributable to the of the treatment effects.
DBT intervention. Future studies using controlled designs with
Our findings indicate that brief DBT skills train- active comparators are needed before strong con-
ing has its impact on the management of acute clusions can be drawn about the suitability of this
symptoms of BPD (e.g. self-destructive behaviour brief intervention for all individuals with BPD. In
and anger), problems that frequently challenge addition, because variability in response is inevita-
healthcare professionals. Although the DBT group ble, studies are needed to identify the factors that
showed superior improvements on healthcare uti- moderate, mediate and predict differential treat-
lization and mental health symptoms (e.g. border- ment outcomes. However, these findings indicate
line symptoms, symptom distress and social that brief DBT skills training has benefits and is not
adjustment) during the treatment period, these harmful for suicidal individuals with BPD. This
gains were not maintained postdischarge. While treatment option should be considered for this high-
our findings contrast with evidence on standard risk population, especially for those unable to access
DBT supporting the maintenance of treatment lengthy comprehensive specialist programmes.
gains postdischarge (e.g. 8, 22, 41), they are consis-
tent with other evidence showing that the effects of Acknowledgements
the skills-only component of DBT (8) is not dur-
able on specific outcomes. Most of the DBT group The authors extend sincere gratitude to all of the patients and
therapists who participated in this trial, without whom this
still reported high levels of general symptom dis-
research would not have been possible. The study was funded
tress at 32 weeks, suggesting that brief DBT skills by the Ontario Mental Health Foundation.
training is not a panacea for a severe patient popu-
lation. However, it may be useful for addressing
specific symptoms in individuals in the acute phase Declaration of interest
of this disorder. Our findings indicate that DBT The authors report no conflict of interests in relationship to
skills training was sufficient for stabilizing high- this study. The clinicaltrials.gov identifier for this study is
risk behaviours and reducing the use of costly crisis NCT01193205, August 31, 2010.
services. Consequently, when long-term compre-
hensive treatment is not accessible, brief DBT References
skills training may be a reasonable intervention to
recommend as a first step, although not as a 1. Grant BF, Chou SP, Goldstein RB et al. Prevalence, corre-
lates, disability, and comorbidity of DSM-IV borderline
replacement for comprehensive treatment. personality disorder: results from the Wave 2 National
Strengths of the present study include its well- Epidemiologic Survey on Alcohol and Related Conditions.
powered, randomized design and its high ratings J Clin Psychiatry 2008;69:533–545.
of treatment adherence. Some limitations need to 2. Lenzenweger MF, Lane MC, Lornager AW, Kessler RC.
DSM-IV personality disorders in the National Comorbid-
be considered. First, the problem with an active
ity Survey Replication. Biol Psychiatry 2007;62:553.
WL control design is that there is no control over 3. Gunderson JG. Borderline personality disorder: a clinical
what the participants do while on the waitlist, guide. Washington DC: American Psychiatric Press, 2001.
such as seeking other forms of treatment. Addi- 4. Gunderson JG. Borderline personality disorder. Washing-
tionally, the absence of a uniform therapeutic ton DC: American Psychiatric Press, 1984.
comparator group prevents us from answering 5. Clarkin JE, Widiger TA, Frances A, Hurt SW, Gilmore M.
Prototypic typology and the borderline personality disor-
questions about the unique effects of DBT skills. der. J Abnorm Psychol 1983;92:263–275.
However, the design is justifiable in the absence of 6. Stoffers JM, Vollm BA, Rucker G, Timmer A, Huband N,
a therapeutic comparator group and an existing Lieb K. Psychological therapies for people with border-
alternative best treatment. The study design can line personality disorder. Cochrane Database Syst Rev
also be seen as a strength because the study more 2012.
7. Soler J, Pascual JC, Tiana T et al. Dialectical behaviour
accurately mirrors the real world. Second, to max- therapy skills training compared to standard group ther-
imize external validity, ancillary treatments were apy in borderline personality disorder: a 3-month ran-
not excluded, so the possibility of confounds must domised controlled clinical trial. Behav Res Ther
be acknowledged. Third, the primary outcome of 2009;47:353–358.
frequency of suicidal and NSSI behaviours was 8. Linehan MM, Korslund KE, Harned MS et al. Dialectical
behavior therapy for high suicide risk in individuals with
self-report (e.g. LSASI; DSHI) and may have borderline personality disorder: a randomized clinical trial
been affected by responder bias. In addition, the and component analysis. JAMA Psychiatry 2015;72:475–
psychometric properties of the LSASI are 482.

147
McMain et al.

9. Gratz KL, Gunderson JG. Preliminary data on an accep- 25. Linehan MM, Heard HL. Treatment history interview
tance-based emotion regulation group intervention for (THI-2). Seattle, WA: University of Washington, 1987.
deliberate self-harm among women with borderline per- 26. Bohus M, Kleindienst N, Limberger MF et al. The short
sonality disorder. Behav Ther 2006;37:25–35. version of the Borderline Symptom List (BSL-23): devel-
10. Blum N, John DS, Pfohl B et al. Systems training for emo- opment and initial data on psychometric properties. Psy-
tional predictability and problem solving (STEPPS) for chopathology 2009;42:32–39.
outpatients with borderline personality disorder: a ran- 27. Spielberger CD. Manual for the state-trait anger expres-
domized controlled trial and 1-year follow-up. Am J Psy- sion inventory. Odessa, FL: Psychological Assessment
chiatry 2008;165:468–478. Resources Inc., 1988.
11. Bos EH, van Wel EB, Appelo MT, Verbraak MJ. A ran- 28. Derogatis LR, Melisaratos N. The brief symptom inven-
domized controlled trial of a Dutch version of systems tory: an introductory report. Psychol Med 1983;13:595–
training for emotional predictability and problem solving 605.
for borderline personality disorder. J Nerv Ment Dis 29. Patton JH, Stanford MS, Barratt ES. Factor structure of
2010;198:299–304. the Barratt Impulsiveness Scale. J Clin Psychol
12. VAN WEL EB, Bos EH, Appelo MT, Berendsen EM, Will- 1995;51:768–774.
geroth FC, Verbraak MJ. The efficacy of the systems train- 30. Beck AT, Steer RA, Brown GK. Manual for beck depres-
ing for emotional predictability and problem solving sion inventory II. San Antonio, TX: Psychological Corpo-
(STEPPS) in the treatment of borderline personality disor- ration, 1996.
der. A randomized controlled trial. Tijdschr Psychiatr 31. Weissman MM, Bothwell S. Assessment of social adjust-
2009;51:291–301. ment by patient self-report. Arch Gen Psychiatry
13. Valentine SE, Bankoff SM, Poulin RM, Reidler EB, Pan- 1976;33:1111–1115.
talone DW. The use of Dialectical Behavior Therapy skills 32. Gratz KL, Roemer L. Multidimensional assessment of
training as stand-alone treatment: a systematic review of emotion regulation and dysregulation: development, fac-
the treatment outcome literature. J Clin Psychol tor structure, and initial validation of the difficulties in
2015;71:1. emotion regulation scale. J Psychopathol Behav Assess
14. AMERICAN PSYCHIATRIC ASSOCIATION. Diagnostic and 2004;26:41–54.
statistical manual of mental disorders, fourth edition 33. Simons JS, Gaher RM. The distress tolerance scale: devel-
(DSM-IV). 4th ed. Washington, DC: American Psychi- opment and validation of a self-report measure. Motiva-
atric Association Press, 1994. tion Emotion 2005;29:83–102.
15. First MB, Spitzer RL, Gibbon M, Williams JBW. Struc- 34. Baer RA, Smith GT, Allen KB. Assessment of mindful-
tured clinical interview for axis I DSM-IV disorders- ness by self-report: the Kentucky inventory of mindfulness
patient edition (SCID-I/P). New York: Biometrics skills. Assessment 2004;11:191–206.
Research Department, NY Sate Psychiatric Institute, 35. Moras KK. Early termination and the outcome of psy-
1995. chotherapy: patient’s perspectives. Dissertation Abstr Int
16. Loranger AW. International personality disorder exami- 1986;46:2817–2818.
nation (IPDE) manual. White Plains, NY: Cornell Medi- 36. Rabe-Hesketh S, Skrondal A. Multilevel and longitudinal
cal Center, 1997. modeling using stata. 2nd ed. Texas: Stata Press, 2008.
17. Linehan MM. Cognitive behavioural treatment of border- 37. Elston DA, Moss R, Boulinier T, Arrowsmith C, Lambin X.
line personality disorder. New York: Guilford Press, Analysis of aggregation, a worked example: number of
1993. ticks on red grouse chicks. Parasitology 2001;122:563–569.
18. Linehan MM. The skills training manual for treating bor- 38. Jacobson NS, Roberts LJ, Berns SB, McGlinchey JB. Meth-
derline personality disorder. New York: Guilford Press, ods for defining and determining the clinical significance
1993. of treatment effects: description, application, and alterna-
19. Linehan MM. DBTÒ skills training manual. New York: tives. J Consult Clin Psychol 1999;67:300–307.
Guilford Publications, 2014. 39. Neacsiu AD, Rizvi SL, Linehan M. Dialectical behavior
20. Gratz KL. Measurement of deliberate self-harm: prelimi- therapy skills use as a mediator and outcome of treatment
nary data on the deliberate self-harm inventory. J Psy- for borderline personality disorder. Behav Res Ther
chopathol Behav Assess 2001;23:253–263. 2010;48:832–839.
21. Linehan MM, Armstrong HE, Suarez A, Allmon D, Heard 40. Gratz KL, Dixon-Gordon KL, Tull MT. Predictors of treat-
HL. Cognitive-behavioral treatment of chronically para- ment response to an adjunctive emotion regulation group
suicidal borderline patients. Arch Gen Psychiatry therapy for deliberate self-harm among women with bor-
1991;48:1060–1064. derline personality disorder. Personal Disord 2014;5:97.
22. Linehan MM, Comtois KA, Murray AM et al. Two-year 41. McMain SF, Guimond T, Streiner DL, Cardish RJ, Links
randomized control trial and follow-up of dialectical PS. Dialectical behavior therapy compared with general
behavior therapy vs. therapy by experts for suicidal behav- psychiatric management for borderline personality disor-
iors and borderline personality disorder. Arch Gen Psychi- der: clinical outcomes and functioning over a 2-year fol-
atry 2006;63:757–766. low-up. Am J Psychiatry 2012;169:650–661.
23. McMain SF, Links PS, Gnam WH et al. A randomized trial
of dialectical behaviour therapy versus general psychiatric
management for borderline personality disorder. Am J Supporting Information
Psychiatry 2009;166:1365–1374.
24. Verheul R. Dialectical behaviour therapy for women with Additional Supporting Information may be found in the online
borderline personality disorder: 12-month, randomised version of this article:
clinical trial in The Netherlands. Br J Psychiatry Fig. S1. Clinically relevant change, by group assignment, at 20
2003;182:135–140. and 32 weeks after randomizationa.

148
Copyright of Acta Psychiatrica Scandinavica is the property of Wiley-Blackwell and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.