ORIGINAL ARTICLE

Two-Year Randomized Controlled Trial

and Follow-up of Dialectical Behavior Therapy
vs Therapy by Experts for Suicidal Behaviors
and Borderline Personality Disorder
Marsha M. Linehan, PhD; Katherine Anne Comtois, PhD; Angela M. Murray, MA, MSW;
Milton Z. Brown, PhD; Robert J. Gallop, PhD; Heidi L. Heard, PhD; Kathryn E. Korslund, PhD;
Darren A. Tutek, MS; Sarah K. Reynolds, PhD; Noam Lindenboim, MS

Context: Dialectical behavior therapy (DBT) is a treat- Main Outcome Measures: Trimester assessments of
ment for suicidal behavior and borderline personality dis- suicidal behaviors, emergency services use, and general
order with well-documented efficacy. psychological functioning. Measures were selected based
on previous outcome studies of DBT. Outcome vari-
Objective: To evaluate the hypothesis that unique as- ables were evaluated by blinded assessors.
pects of DBT are more efficacious compared with treat-
ment offered by non–behavioral psychotherapy experts. Results: Dialectical behavior therapy was associated with
better outcomes in the intent-to-treat analysis than com-
Design: One-year randomized controlled trial, plus 1 munity treatment by experts in most target areas during
year of posttreatment follow-up. the 2-year treatment and follow-up period. Subjects re-
ceiving DBT were half as likely to make a suicide at-
Setting: University outpatient clinic and community
tempt (hazard ratio, 2.66; P=.005), required less hospi-
talization for suicide ideation (F1,92=7.3; P=.004), and had
practice.
lower medical risk (F1,50=3.2; P=.04) across all suicide
attempts and self-injurious acts combined. Subjects re-
Participants: One hundred one clinically referred
ceiving DBT were less likely to drop out of treatment (haz-
women with recent suicidal and self-injurious behav- ard ratio, 3.2; P⬍.001) and had fewer psychiatric hos-
iors meeting DSM-IV criteria, matched to condition on pitalizations (F1,92=6.0; P=.007) and psychiatric emergency
age, suicide attempt history, negative prognostic indica- department visits (F1,92=2.9; P =.04).
tion, and number of lifetime intentional self-injuries and
psychiatric hospitalizations. Conclusions: Our findings replicate those of previous stud-
ies of DBT and suggest that the effectiveness of DBT can-
Intervention: One year of DBT or 1 year of commu- not reasonably be attributed to general factors associated
nity treatment by experts (developed to maximize inter- with expert psychotherapy. Dialectical behavior therapy
nal validity by controlling for therapist sex, availability, appearstobeuniquelyeffectiveinreducingsuicideattempts.
expertise, allegiance, training and experience, consulta-
tion availability, and institutional prestige). Arch Gen Psychiatry. 2006;63:757-766

S
UICIDAL BEHAVIOR IS A BROAD ior,4-9 with a suicide rate of up to 9%.10
term that includes death by Forty percent of the highest users of in-
suicide and intentional, non- patient psychiatric services receive a di-
fatal, self-injurious acts com- agnosis of BPD.11,12 Patients with BPD use
mitted with or without in- more services than those with major de-
tent to die. It is associated with several pression13 and other personality disor-
mental disorders, including depression, ders.14 Among patients with BPD seen for
substance dependence, and schizophre- treatment, 72% have had at least 1 psy-
nia. Borderline personality disorder (BPD) chiatric hospitalization and 97% have re-
is 1 of only 2 DSM-IV diagnoses for which ceived outpatient treatment from a mean
suicidal behavior is a criterion.1 Border- of 6.1 previous therapists.15,16 Despite this
line personality disorder is a severe and high-use pattern, patients with BPD have
persistent mental disorder experience of high rates of treatment failure.17,18
severe emotional distress and behavioral Outpatient dialectical behavior therapy
Author Affiliations are listed at dyscontrol.1-3 Among patients with BPD, (DBT)20,21 and mentalization-based treat-
the end of this article. 69% to 80% engage in suicidal behav- ment provided in a partial hospital pro-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

757

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 interfering (dropout from treatment) behaviors and in
Enrollment
186 Assessed for Eligibility promoting quality-of-life behaviors (use of hospital ser-
vices for suicidal behaviors and psychological issues).
75 Excluded
53 Did Not Meet Inclusion
Criteria METHODS
22 Refused to Participate

111 Randomized PARTICIPANTS AND ASSESSMENTS

Allocation
60 Allocated to DBT 51 Allocated to CTBE
Participants were women between the ages of 18 and 45 years
52 Received DBT 49 Received CTBE who met criteria for BPD and for current and past suicidal be-
8 Training Cases 2 Pilot Cases havior as defined by at least 2 suicide attempts or self-injuries
Follow-up
in the past 5 years, with at least 1 in the past 8 weeks. Indi-
6 Lost to Follow-up 14 Lost to Follow-up
10 Discontinued Interventions 21 Discontinued Interventions
viduals were excluded if they had (1) a lifetime diagnosis of
schizophrenia, schizoaffective disorder, bipolar disorder, psy-
Analysis
52 Analyzed 49 Analyzed chotic disorder not otherwise specified, or mental retardation;
(2) a seizure disorder requiring medication; (3) a mandate to
treatment; or (4) the need for primary treatment for another
Figure 1. Subject flowchart. CTBE indicates community treatment by debilitating condition. All participants provided informed con-
experts; DBT, dialectical behavior therapy. Those in DBT lost to follow-up sent using protocols approved by the University of Washing-
and discontinued interventions were not subtracted from the allocation total ton Human Subjects Division.
in the DBT treatment arm.
Using a computerized adaptive minimization randomiza-
tion procedure, eligible subjects were matched to treatment con-
dition on 5 primary prognostic variables: (1 and 2) the num-
gram22,23 are the only 2 treatments (to our knowledge) for ber of lifetime suicide attempts or nonsuicidal self-injuries
which randomized controlled trials have been published combined and psychiatric hospitalizations; (3) a history of only
that demonstrate significantly better results than treat- suicide attempts, only nonsuicidal self-injury, or both; (4) age;
ment as usual (TAU) for patients with BPD.24 Of the 2 treat- and (5) a negative prognostic indicator of a Beck Depression
ments, DBT has the most empirical support, having been Inventory34 score higher than 30 or a Global Assessment of Func-
evaluated initially in a 1991 randomized trial25 and follow- tioning35 score lower than 45 for a comorbid condition. This
up19,26,27 and assessed subsequently in 4 published random- matching method has been shown to be superior to simple and
ized controlled trials28-31 across 3 separate research cen- stratified randomization in producing balance for separate prog-
nostic variables, particularly when the number of strata is large
ters and in 1 additional randomized controlled trial of a compared with the number of subjects.36 Based on 0.8 power
DBT-oriented treatment.32 Evidence from these studies has to detect significant differences between conditions (P = .05,
shown DBT to be an efficacious treatment for BPD. How- 1-sided), this procedure was used to randomize 101 subjects
ever, as noted in the American Psychiatric Association Prac- to DBT (n=52) or to CTBE (n=49). The flow of subjects through
tice Guidelines, “[I]t is difficult to ascertain whether the the study is shown in Figure 1.
improvement reported for patients receiving dialectical be- Initial assessments were done before informing subjects of
havior therapy derived from specific ingredients of dialec- treatment assignment and at 4-month intervals during the treat-
tical behavior therapy.”33(p32) The present study is the first ment and follow-up periods. Outcome assessments were yoked
in a program of research systematically examining DBT with across conditions by screening date. Assessments were con-
the objective of identifying the specific elements of treat- ducted by blinded independent clinical assessors with mas-
ter’s or doctoral degrees. Lead assessors (K.A.C. and A.M.M.)
ment that are necessary and sufficient for an efficacious out- were trained on interview measures by the instrument devel-
come among individuals with BPD. As such, this is not a opers or by an approved trainer, and then trained, supervised,
“horse race” study pitting one complex active treatment and evaluated for reliability across assessors (␬ statistic or in-
against another. Rather, it is a dismantling study designed traclass correlation coefficient for all ratings ranged from 0.74
to begin answering questions as to the unique effects of DBT. to 1.00). The participant coordinator, who was not blinded to
To rule out factors commonly believed to be effective treatment condition, executed the randomization program and
across a variety of disorders, the control condition, com- collected all the data related to treatment.
munity treatment by experts (CTBE), was specifically de-
signed to maximize internal validity by controlling for Diagnostic Interviews
the following: (1) availability of treatment; (2) assis-
tance finding and getting to a first appointment with a The structured clinical interviews for Axis I and Axis II DSM-
therapist; (3) hours of individual psychotherapy of- IV36,37 and the International Personality Disorder Examina-
fered; (4) therapist sex, training, clinical experience, and tion38 were used as screening and diagnostic instruments. The
Peabody Picture Vocabulary Test–Revised39 was used to rule
expertise; (5) availability of group clinical consultation; out mental retardation.
(6) allegiance to treatment approach; (7) institutional pres-
tige associated with treatment; and (8) general factors as- Outcome Measures
sociated with receiving any psychotherapy. This study
was designed as a replication of the 1991 study25 com- The Suicide Attempt Self-Injury Interview40 measured the to-
paring DBT with TAU. We predicted that we would find pography, suicide intent, and medical severity of each suicide
identical outcomes, namely, that DBT would perform sig- attempt and nonsuicidal self-injury. Interrater reliabilities were
nificantly better than CTBE in reducing suicidal (sui- 0.88 for medical risk and 0.94 for suicide intent. The Suicidal
cide attempts and nonsuicidal self-injury) and therapy- Behaviors Questionnaire (M.M.L., unpublished work, 1981) was

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

758

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 Table 1. Characteristics of Therapists for the Dialectical Behavior Therapy (DBT) and Community Treatment by Experts (CTBE) Groups*

DBT Therapists CTBE Therapists

Characteristic (n = 16) (n = 25) ␹2 or z P Value
Male sex 5 (31.3) 9 (36.0) −0.31 .75
⬎10 y Clinical experience since terminal degree 4 (25.0) 14 (56.0) Fisher exact .06
Doctoral degree 12 (75.0) 14 (56.0) Fisher exact .32
No. of study clients, mean ± SD 3.6 ± 2.9 2.5 ± 1.7 −1.33 .18
Subjects in group clinical consultation† 52 (100.0) 28 (57.1) Fisher exact ⬍.001

*Data are given as number (percentage) unless otherwise indicated.

†Percentage of clients whose therapist had group clinical consultation for that client. The expert consultant for the CTBE condition attended the group clinical
consultation, but the expert consultant for the DBT condition (M.M.L.) did not.

used to assess suicide ideation. The Reasons for Living Inven- tice. The treatment developer (M.M.L.) was not a study thera-
tory41 assessed the importance of reasons for living. The Treat- pist and did not attend the weekly DBT team meetings.
ment History Interview (M.M.L., unpublished work, 1987) mea- Individual therapists were hired once 6 of 8 consecutive train-
sured subjects’ experience with professional psychotherapy, ing case sessions were rated as adherent to DBT. During the study,
comprehensive treatment programs, case management, inpa- adherence was assessed by coding a random selection of ses-
tient units, emergency treatment and other crisis services, and sions on the DBT Global Rating Scale (M.M.L., unpublished work,
medication use. The Hamilton Rating Scale for Depression– 2003), which codes DBT adherence on a 5-point scale (to 1 deci-
17-Item42 was used to evaluate the severity of depressive symp- mal point), with a score of 4.0 or higher denoting adherence.
toms. All outcome domains are identical to those reported by Therapists were blinded to which sessions were rated. Coders
Linehan et al25 in 1991. were trained to reliability with the treatment developer.

STUDY TREATMENTS Community Treatment by Experts

Therapists The CTBE condition was developed specifically for this study to
control for factors previously uncontrolled for in DBT studies.
There were 41 therapists in the study (16 DBT therapists and Similar to a TAU (treatment as usual) condition, the treatment
25 CTBE therapists). Characteristics of the therapists are given provided was uncontrolled by the research team, an essential com-
in Table 1. ponent of TAU conditions. It differs from TAU conditions in that
characteristics of CTBE therapists are controlled by the study via
Dialectical Behavior Therapy selection of therapists and supervisory arrangements.

Dialectical behavior therapy is a cognitive behavioral treat- Expertise. The CTBE therapists were nominated by commu-
ment program developed to treat suicidal clients meeting cri- nity mental health leaders. These included heads of inpatient
teria for BPD.20,21 It directly targets (1) suicidal behavior, (2) psychiatric units and clinical directors of mental health agen-
behaviors that interfere with treatment delivery, and (3) other cies, who nominated therapists whom they considered ex-
dangerous, severe, or destabilizing behaviors. perts in treating difficult clients.
Standard DBT addresses the following 5 functions: (1) in-
creasing behavioral capabilities, (2) improving motivation for Allegiance to Treatment Provided. The content of the treat-
skillful behavior (through contingency management and re- ment provided by the CTBE therapists was not prescribed by
duction of interfering emotions and cognitions), (3) assuring the research study or interfered with in any way. Therapists were
generalization of gains to the natural environment, (4) struc- asked to provide the type and dose of therapy that they be-
turing the treatment environment so that it reinforces func- lieved was most suited to the patient, with a minimum of 1 sched-
tional rather than dysfunctional behaviors, and (5) enhancing uled individual session per week. Ancillary treatment could be
therapist capabilities and motivation to treat patients effec- prescribed as needed.
tively. These functions are divided among the following 4 modes
of service delivery: (1) weekly individual psychotherapy Availability of Clinical Supervision Group. The CTBE thera-
(1 h/wk), (2) group skills training (2½ h/wk), (3) telephone pists were paid at the same rate as that paid to DBT therapists.
consultation (as needed within the therapist’s limits to ensure The CTBE therapists were not required to attend a weekly clini-
generalization), and (4) weekly therapist consultation team meet- cal supervision group.
ings (to enhance therapist motivation and skills and to pro-
vide therapy for the therapists). Institutional Prestige. The CTBE clinical supervision group met
at the Seattle Psychoanalytic Society and Institute and was led
Recruitment, Training, by its training director. The institute’s prestige outside the field
and Adherence of DBT Therapists of behavior therapy in Seattle rivals that of the University of
Washington within the field of behavior therapy.
Psychotherapists recommended by colleagues as potentially good
DBT therapists were recruited for the study; 8 had no previous General Factors and Assistance Finding a Therapist. The par-
DBT exposure and 8 had experience that ranged from work- ticipant coordinator established an independent relationship
shop attendance to applied practice. The sample included 3 with subjects in both conditions. The participant coordinator
graduate students and 2 postdoctoral trainees. Training con- also provided assistance in contacting the therapist and in get-
sisted of a 45-hour DBT seminar followed by supervised prac- ting the subject to the first session.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

759

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 Availability of Affordable and Sufficient Treatment Hours. The With mixed-effects models, appropriate covariance struc-
study paid for CTBE at the same rate as for DBT. All partici- tures were analytically determined. Determination of the ap-
pants paid a small sliding fee for therapy as determined by the propriate covariance structure was based on a mixture of ␹2 tests
assessment team before condition assignment. No partici- in comparing nested models.52
pants were dropped because of failure to pay.
RESULTS
Sex, Training, and Clinical Experience. The randomization
program assigned clients to DBT and CTBE therapists, match-
ing on sex, doctoral vs master’s training, and years of clinical SAMPLE CHARACTERISTICS
experience. Results indicated that therapists’ sex and training AND BASELINE DIFFERENCES
did not differ in the 2 conditions. The CTBE therapists, how-
ever, had more clinical experience, which was expected be- The treatment groups did not significantly differ on demo-
cause they were selected for their expertise (Table 1). graphic characteristics, diagnoses, or pretreatment num-
ber of suicide attempts or nonsuicidal self-injuries. These
Recruiting of CTBE Therapists results are summarized in Table 2.
Ninety-four potential therapists were recruited. Based on their de- STUDY IMPLEMENTATION
scriptions of the treatment that they usually provide to patients
with BPD, therapists were categorized into 6 groups anchored by
Dialectical behavior therapy adherence was rated for 51
“behavior therapist” on one end and “very nonbehavioral” on the
other end. The CTBE therapist pool consisted of 38 therapists self- client-therapist dyads across 571 therapy sessions. Ad-
described as “eclectic but nonbehavioral” or “mostly psychody- herence scores ranged from 2.5 to 4.8, with a mean ± SD
namic.” No cognitive behavior therapists were selected for the score of 4.0±0.2.
CTBE condition. Of the potential 38 CTBE therapists, 25 were Compared with the CTBE group, the DBT group re-
assigned patients (the remaining 13 potential therapists de- ceived significantly more therapy from study therapists, pri-
clined to accept a patient when one was assigned). marily because of their weekly group sessions and their
greater treatment retention (Table 3). There were no sig-
STATISTICAL ANALYSIS nificant differences in total hours of therapy, however, when
all study-provided and non–study-provided treatment hours
Our primary method for analyzing repeated-measures data (group, individual, case management, day treatment, and
(pretreatment through the 24-month assessment) was mixed- inpatient treatment) were summed. There were no differ-
effects models.43 Two versions of mixed-effects models were ences between conditions in the use or the types of psy-
implemented: (1) random regression modeling (RRM), also chotropic medications at pretreatment. During the treat-
known as hierarchical linear models and multilevel linear ment year, the use of psychotropic medications decreased
models,44,45 and (2) mixed-model analysis of variance significantly more in the DBT group than in the CTBE group
(MMANOVA).46 Differences in rates of change (ie, slopes) (t1 =2.6, P⬍.01, RRM). There were no significant differ-
were compared for the 2 treatment groups. Outcome variables ences between conditions during the follow-up year
with nonnormal distributions were recoded into discrete or-
dinal levels and were analyzed using RRM for ordinal data (eg,
(Figure 2). At the pretreatment and 4-month assess-
total nonsuicidal self-injurious acts were recoded into 5 ordi- ments, the DBT and CTBE groups did not significantly dif-
nal levels).47,48 fer in their expectations of receiving help from the study
Outcomes that were infrequent (and highly skewed) were therapy. The mean ± SD expectancy ratings were 5.52±0.95
analyzed by means of a generalized MMANOVA using all cli- and 5.42 ± 1.02 at the pretreatment assessment and
ents with any outcome data (n=97), with the pretreatment val- 5.46±0.98 and 5.29±1.37 at the 4-month assessment, for
ues as a covariate. The MMANOVA required that suicide at- the DBT and CTBE groups, respectively.
tempts, emergency department visits, and inpatient admissions
be recoded into binary variables (none vs any) per assessment ANALYSIS OF THE EFFECT
period; therefore, absolute frequency of these outcomes could OF MISSING DATA PATTERNS
not be statistically compared.49
Cox proportional hazards regression model survival analy-
sis tested the time to first suicide attempt.50 For simple cross-
Eight episodes were not included in the analyses of suicide
sectional comparisons (eg, pretreatment differences and ma- attempts and nonsuicidal self-injury because the subjects
nipulation checks), t tests were used for normally distributed didnotprovideenoughinformationintheinterviewstomake
variables, Mann-Whitney tests for nonnormal variables, and ␹2 that classification. The information was missing because the
tests for binary variables. subjects refused to answer the question or said they could
To assess the potential effect of missing data (ie, ignorable not remember, or the interviewer skipped needed questions
vs informative missing data), a pattern-mixture analysis was because the subject was too emotionally distraught.
implemented using 2-tailed tests.51 We defined patterns using Compared with subjects assigned to the DBT condi-
a binary completer status variable, which was entered as a pre- tion, subjects assigned to the CTBE condition were sig-
dictor in the RRM and MMANOVA. To determine if the slope nificantly more likely to drop out of study assessments
differences depend on completer status, a 3-way interaction of
completer status–⫻–treatment group–⫻-time was included in
during the treatment year (3.8% vs 20.4%, P=.01, Fisher
the hierarchical linear models. To determine if the mean dif- exact test) and during the overall study (11.5% vs 28.6%,
ference between groups depends on completer status, a 2-way P=.03, Fisher exact test). We examined the effects of dif-
interaction of completer status–⫻–treatment group was in- ferential missing data and treatment dropout on each of
cluded in the MMANOVA. The same analyses were con- our major outcome variables and found no evidence that
ducted for treatment dropout status. the findings were biased by these differences.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

760

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 Table 2. Pretreatment Demographic, Suicide Attempt, Nonsuicidal Self-injury, and Diagnostic Data for the Dialectical Behavior
Therapy (DBT) and Community Treatment by Experts (CTBE) Groups*

DBT Group CTBE Group

Variable (n = 52) (n = 49) Total
Demographic Characteristics
Age, mean ± SD, y 29.0 ± 7.3 29.6 ± 7.8 29.3 ± 7.5
Race
African American 3.8 4.1 4.0
Asian American 1.9 2.0 2.0
Native American or Alaskan Native 1.9 0 1.0
Other 5.8 6.1 5.0
White 86.5 87.8 87
Single, divorced, or separated 88.4 85.7 87.2
Education
⬍High school 9.6 6.1 7.9
High school graduate or general equivalency diploma 15.4 18.4 16.8
Some college or technical school 50.0 53.1 51.5
College graduate 25.0 22.4 23.8
Annual income, $
⬍15 000 75.0 75.5 75.2
15 000-30 000 13.4 8.2 10.9
⬎30 000-50 000 9.6 10.2 9.9
Suicide Attempt and Nonsuicidal Self-injury, Median (Interquartile Range)
Suicide attempt 1.0 (0.0-3.0) 2.0 (1.0-4.0) 1.0 (0.5-4.0)
Nonsuicidal self-injury 11.0 (2.3-51.5) 7.0 (2.0-34.0) 10.0 (2.0-47.0)
Diagnostic Data
Lifetime psychiatric diagnoses meeting DSM-IV criteria†
Major depressive disorder 94.2 98.0 96.0
Panic disorder 48.1 55.1 51.5
Posttraumatic stress disorder 59.6 51.0 55.4
Any anxiety disorder 90.4 83.7 87.1
Any substance use disorder 78.8 67.3 73.3
Any eating disorder 44.2 34.7 39.6
Current psychiatric diagnoses meeting DSM-IV criteria†
Major depressive disorder 71.2 73.5 72.3
Panic disorder 42.3 38.8 40.6
Posttraumatic stress disorder 50.0 49.0 49.5
Any anxiety disorder 82.7 73.5 78.2
Any substance use disorder 23.0 36.7 29.7
Any eating disorder 25.0 22.4 23.8
Axis II
Cluster A 1.9 4.1 3.0
Cluster B other than borderline personality disorder 11.5 10.2 10.9
Cluster C 25.0 26.5 25.7
Paranoid 1.9 4.1 3.0
Schizoid 0.0 0.0 0.0
Schizotypal 0.0 0.0 0.0
Antisocial 11.5 10.2 10.9
Histrionic 1.9 2.0 2.0
Narcissistic 0.0 0.0 0.0
Avoidant 17.3 24.5 20.8
Dependent 3.8 8.2 5.9
Obsessive compulsive 7.7 8.2 7.9
Psychiatric disorder not otherwise specified 94.2 83.7 89.1

*Data are given as percentages unless otherwise indicated. No values were statistically significant. Analyses were conducted using the t test, Mann-Whitney
test, and ␹2 test as appropriate.
†The first 8 subjects entering the study met criteria using the DSM-III-R.

SUICIDAL BEHAVIORS terval {CI}, 2.40-18.07]) (Figure 3). The NNT of 4.24 in-
dicates that, during 2 years of treatment plus follow-up, 4
There were no documented suicides in either condition dur- patients would need to be treated with DBT to prevent 1
ing the 2-year study. The DBT group had half the rate of patient from attempting suicide. Similarly, half as many sub-
suicide attempts compared with the CTBE group (23.1% jectsintheDBTgroupmadenonambivalentsuicideattempts
vs 46%, ␹12 =5.98, P=.01; hazard ratio, 2.66, P=.005; and (5.8%vs13.3%,P=.18,FisherexacttestandNNT,13.3[95%
number needed to treat [NNT], 4.24 [95% confidence in- CI, 5.28-25.41]). There were significantly fewer suicide at-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

761

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 Table 3. Treatment Implementation and Treatment Dropouts for the Dialectical Behavior Therapy (DBT) and Community Treatment
by Experts (CTBE) Groups*

DBT Group CTBE Group

Variable (n = 52) (n = 49)
Treatment Implementation
First year
No. of individual study therapy sessions, study† 42.5 (33.3-51.5) 33.0 (14.0-46.0)‡
No. of all individual therapy sessions§ 51.0 (42.5-57.0) 45.0 (21.0-62.5)
No. of group therapy sessions, study† 38.0 (25.3-45.3) 0.0 (0.0-0.0) 㛳
No. of all group therapy session§ 38.0 (26.0-45.3) 0.0 (0.0-0.0) 㛳
Total No. of treatment hours¶ 64.7 (56.0-73.7) 56.0 (21.0-86.9)
No. of weeks in study treatment# 52.5 (50.0-73.7) 50.9 (20.8-53.0) 㛳
Second year
Any outpatient therapy, % 71.2 67.6
Individual therapy, % 63.0 67.6
No. of individual therapy sessions 6.5 (0.0-28.0) 9.0 (0.0-37.5)
Total No. of treatment hours 16.8 (0.0-42.0) 17.0 (0.0-64.0)
Treatment Dropouts
Dropped first study therapist, No. (%) 13 (25.0) 29 (59.2) 㛳
No. of weeks before dropped first therapist 16.9 (7.0-31.4) 9.7 (2.4-26.3)
No. who saw 2 therapists, No. (%) 5 (9.6) 12 (24.5)‡
No. who saw 3 therapists, No. (%) 0 2 (4.1)

*Data are given as median (interquartile range) unless otherwise indicated. Proportions were compared using ␹2 tests, and continuous variables were
compared using the Mann-Whitney test. P values are 2-tailed.
†Individual therapy sessions by study therapists during the treatment year.
‡P⬍.05.
§Any therapy sessions during the year, including sessions outside the study.
㛳P⬍.001.
¶Total inpatient and outpatient treatment time. Each session of individual therapy, family therapy, and vocational counseling was counted as 1 hour of therapy;
each group therapy session was counted as 20 minutes of therapy; each day of day treatment was counted as 30 minutes of therapy; and each psychiatric
inpatient day was counted as 3½ hours of therapy.
#Total number of weeks clients saw any study therapist, including time after dropping first therapist.

pretreatment year (F1,94 =3.20, P=.04, MMANOVA). The

1.0 mean proportions of suicide attempters per treatment group
perperiodwere6.2%(95%CI,3.1%-11.7%)and12.2%(95%
CI,7.1%-20.3%)fortheDBTandCTBEgroups,respectively.
0.9 Across the 2 years, the median number of suicide attempts
Proportion Taking Any Psychotropic Medication

was 0 for the DBT group and for the CTBE group. The in-
terquartile range for suicide attempts during the 2 years was
0.8 0 to 0 for the DBT group and 0 to 1 for the CTBE group.
Similarly,bothtreatmentswereeffectiveinreducingthenum-
ber of nonsuicidal self-injuries (␹12 =120.6, P⬍.001, ordi-
0.7 nal RRM), but the difference in the rates of change was not
significant (F1,99 =1.1, P=.15 [standardized effect size, 0.47])
(Figure 4). The median number of nonsuicidal acts for
0.6 the 2 years was 3.0 (interquartile range, 1.0-7.8) for the DBT
group and 3.0 (interquartile range, 0.0-8.0) for the CTBE
group.
0.5 As summarized in Table 4, among subjects with any
Treatment Group suicide attempt or intentional self-injury during the treat-
DBT
TBE ment year, the highest medical risk was significantly lower
0.4
for the DBT group than for the CTBE group (F1,156 =3.2,
Pretreatment 4 8 12 16 20 24 P=.04). Both treatment groups made significant improve-
Assessment Period, mo ment in suicide ideation and in reasons for living (P⬍.001
for both), but the slope difference between conditions
Figure 2. Proportion of subjects taking any psychotropic medication in the was not significant.
past 2 months. The treatment period ended at 12 months, and the follow-up
period ended at 24 months. CTBE indicates community treatment by experts;
DBT, dialectical behavior therapy. USE OF CRISIS SERVICES

tempts per period in the DBT group across the 2 years when Based on results of the MMANOVA, the DBT group used
controlling for the number of suicide attempts during the crisisservicessignificantlylessthantheCTBEgroupthrough-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

762

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 out the 2 years of the study (Table 5). Fewer DBT subjects
went to the emergency department at least once for any psy- 1.0

chiatric reason, including drug or alcohol problems (year

1: 43.1% of DBT subjects vs 57.8% of CTBE subjects, and
0.9
year 2: 23.4% of DBT subjects vs 28.9% of CTBE subjects;
F1,92 =2.9, P=.04; NNT, 9.09 [95% CI, 3.30-12.04]), or for
suicide ideation specifically (year 1: 15.7% of DBT subjects
0.8
had at least 1 visit vs 33.3% of CTBE subjects, and year 2:

Proportion Not Attempting Suicide

10.6% of DBT subjects vs 18.4% of CTBE subjects; F1,92 =4.3,
P=.02; NNT, 4.42 [95% CI, 2.49-19.76]). 0.7
In addition, significantly fewer DBT subjects were ad-
mitted to hospitals for any psychiatric reason (year 1:
19.6% of DBT subjects had at least 1 admission vs 48.9% 0.6
of CTBE subjects, and year 2: 23.4% of DBT subjects vs
Treatment Group
23.7% of CTBE subjects; F1,92 = 6.0, P = .007; NNT, 3.88 DBT
[95% CI, 2.26-13.71]) or for suicide ideation specifi- 0.5 CTBE
cally (year 1: 9.8% of DBT subjects had at least 1 admis- DBT Censored
CTBE Censored
sion vs 35.6% of CTBE subjects, and year 2: 14.9% of DBT
0.4
subjects vs 18.4% of CTBE subjects; F1,92 = 7.3, P =.004; 0 200 400 600 800
NNT, 4.46 [95% CI, 2.53-19.17]). Time to First Suicide Attempt, d

BEHAVIORS THAT INTERFERE WITH THERAPY Figure 3. Survival analysis for time to first suicide attempt. The treatment
period ended at 365 days, and the follow-up period ended at 730 days. CTBE
indicates community treatment by experts; DBT, dialectical behavior therapy.
More CTBE than DBT subjects dropped out of the study
therapy (Table 3). Subjects could choose reassignment
to up to 2 additional therapists for the same condition.
2.50
Cox proportional hazards regression model survival analy-
sis indicated that the risk of dropping out of therapy was
3 times higher for CTBE subjects for dropping the first
therapist (hazard ratio, 3.2, P⬍.001; and NNT, 2.92 [95% 2.00
CI, 1.91-6.21]) and for dropping therapy entirely (rela-
tive risk ratio, 2.7, P=.01; and NNT, 4.22 [95% CI, 2.43-
Mean Ordinal Nonsuicidal Self-injury

16.16]). In all, 29 CTBE subjects (59.2%) and 13 DBT

subjects (25.0%) dropped their first study therapist. 1.50
Twenty-one CTBE subjects (42.9%) and 10 DBT sub-
jects (19.2%) dropped all study therapy (P = .005).
1.00
BEHAVIORS REFLECTING QUALITY OF LIFE

Both treatment groups had significant reductions in scores

on the Hamilton Rating Scale for Depression–17 Item 0.50
(P⬍.001 for both). However, the slope difference be-
Treatment Group
tween the treatment groups was not significant (Table 4). DBT
CTBE

0.00
COMMENT Pretreatment 4 8 12 16 20 24

Assessment Period, mo
This study compared DBT with a rigorous comparison con-
dition, nonbehavioral CTBE, to address whether the effec- Figure 4. Mean ordinal nonsuicidal self-injury during the 2-year study.1 The
tiveness of DBT in treating suicidal patients and patients treatment period ended at 12 months, and the follow-up period ended at 24
with BPD can be accounted for by treatment factors com- months. The 5-level ordinal categories per assessment period were 0, 0.01
to 1, 1.01 to 2, 2.01 to 4, and 4.01 and higher. CTBE indicates community
mon to most psychotherapy by experts. Results indicated treatment by experts; DBT, dialectical behavior therapy.
that DBT was superior to CTBE in preventing suicide at-
tempts, with a hazard ratio suggesting that suicide at-
tempts can be reduced by half with DBT compared with findings of this study indicate that the efficacy of DBT can-
non–behavioral therapy by experts. Dialectical behavior not reasonably be attributed solely to general factors asso-
therapy was also more effective in reducing emergency de- ciated with receiving expert psychotherapy.
partment visits and inpatient psychiatric care for suicide The hazard ratio for nonambivalent suicide attempts (ie,
ideation. In addition, DBT was more than twice as effec- those with high intent and planning) was 2.2, almost iden-
tive as non–behavioral therapy by experts in keeping sub- tical to that for suicide attempts overall. (We cannot con-
jects in treatment, as reflected by a 25% dropout rate from clude that this is a statistically significant difference, as the
the first therapist in DBT compared with 59% in CTBE. The low base rate of serious suicide attempts precludes adequate

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

763

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 Table 4. Longitudinal Outcome Measures for the Dialectical Behavior Therapy (DBT)
and Community Treatment by Experts (CTBE) Groups*

Pretreatment 12-mo Posttreatment 24-mo Follow-up ⌬ in Slopes

DBT Group CTBE Group DBT Group CTBE Group DBT Group CTBE Group P
Variable (n = 52) (n = 49) (n = 50) (n = 39) (n = 46) (n = 35) F Value
Highest medical risk† 7.1 ± 4.9 8.8 ± 4.9 5.0 ± 4.2 7.4 ± 5.6 ... ... 3.2 .04
Suicide ideation 51.7 ± 20.3 59.9 ± 21.6 29.8 ± 24.5 32.8 ± 26.3 24.1 ± 19.8 31.92 ± 26.8 0.2 .31
Reasons for Living Inventory
Mean total item score 2.8 ± 0.7 2.7 ± 0.9 3.2 ± 0.9 3.0 ± 0.8 3.3 ± 0.9 3.1 ± 0.8 0.9 .17
Survival and coping 2.7 ± 0.9 2.7 ± 1.0 3.4 ± 1.2 3.3 ± 1.2 3.7 ± 1.0 3.3 ± 1.4 1.4 .12
Hamilton Rating Scale for 20.2 ± 5.9 21.7 ± 7.3 14.0 ± 7.3 17.0 ± 8.2 12.6 ± 6.8 14.4 ± 9.1 0.0 .43
Depression–17 Item

*Data are given as mean±SD unless otherwise indicated. Reported means are estimates of the random regression modeling (RRM). Unless otherwise
specified, slope and intercept were included as random effects in standard linear RRM, and RRM was based on all available data (pretreatment, 4-, 8-, 12-, 16-,
20-, and 24-month assessments).
†Analysis of combined suicide attempt and self-injury data aggregated per year includes only subjects with suicide attempt or nonsuicidal self-injury during the
treatment year. There were too few acts during the follow-up year for analysis. Random intercept RRM (without slope as a random effect) was used.

Table 5. Emergency Department Visits and Hospital Admissions for the Dialectical Behavior Therapy (DBT)
and Community Treatment by Experts (CTBE) Groups*

Treatment Year Follow-up Year

DBT Group CTBE Group DBT Group CTBE Group

Median Median Median Median

(Interquartile (Interquartile (Interquartile (Interquartile P
Variable % Range) % Range) % Range) % Range) F Value
Emergency Department Visits
For psychiatric reason 43.1 0 (0-1) 57.8 1 (0.0-2.5) 23.4 0 (0-0) 28.9 0 (0.0-1.0) 2.9 .04
For suicide ideation 15.7 0 (0-0) 33.3 0 (0.0-1.0) 10.6 0 (0-0) 18.4 0 (0.0-0.0) 4.3 .02
Hospital Admissions
For psychiatric reason 19.6 0 (0-0) 48.9 0 (0.0-2.0) 23.4 0 (0-0) 23.7 0 (0.0-0.3) 6.0 .007
For suicide ideation 9.8 0 (0-0) 35.6 0 (0.0-1.0) 14.9 0 (0-0) 18.4 0 (0.0-0.0) 7.3 .004

*Differences were tested by means of mixed-model analysis of variance using data from all assessment periods (pretreatment, 4-, 8-, 12-, 16-, 20-, and
24-month assessments). The highly skewed distribution analysis precluded a statistical comparison of absolute numbers of these events.

power in any single-site clinical trial.) This finding, how- (16.82±60.81) for subjects assigned to CTBE in the present
ever, combined with the significantly lower risk for any type study; in contrast, DBT had outcomes in the present study
of suicide attempt suggests that DBT may be uniquely ef- almostidenticaltothosefoundinthe1991study(6.05±11.55
fective in treating suicidal individuals. Similar to other DBT and 6.38±7.41 for the 1991 study and the present study, re-
randomized trials,26,28-32,53 there was a low mortality rate dur- spectively). The differences between conditions in outcome
ing the study. There were no documented suicides (other variability across the 2 studies are remarkably similar, with
than 1 death in the CTBE group related to the cumulative DBT outcomes being consistently less variable across per-
effectsofprevioussuicideattempts).Theabsenceofanydeath sons than those of either comparison condition. It is pos-
bysuicidemaybeduetoanynumberoffactors.Subjectswere sible that the present study may have been underpowered
offered not only expert therapy but also extensive contact to detect the 0.49 effect size of DBT vs CTBE.
with an assessment team. They were mailed frequent non- An alternate explanation of our findings may be that sub-
demanding cards throughout the treatment and follow-up jects in CTBE underreported habitual self-injury to a greater
years. Such a regimen has been found to reduce completed extent than those in DBT. In a previous study30 among drug
suicides among individuals treated for suicidality.54 abusers, retrospective reports of opiate use at 4-month in-
In contrast to previous DBT randomized trials,25,28,31,32,54 tervals to a blinded assessor were compared with contem-
there were no significant differences between conditions in poraneous urinalyses data. The correlation of DBT subjects’
the incidence or the frequency of nonsuicidal self-injury in self-reports with urinalyses data was 0.71, whereas the cor-
our study. An examination of outcomes from the 1991 DBT relation in the nonbehavioral control condition was 0.02,
trialofsuicidalwomenwithBPDsuggeststhatnon–behavioral despite a lack of any negative consequence for self-reporting
CTBE may be more effective than TAU in reducing nonsui- drug use. Subjects in DBT self-reported using opiates more
cidal self-injury. The mean±SD number of nonsuicidal acts frequentlyevenwhenthrice-weeklyurinalysesindicatedthat
during the treatment year for subjects assigned to TAU in they were actually using opiates less frequently.30 Daily di-
the 1991 study25 was 32.32±69.97, with half that many ary keeping and weekly discussion of recorded behaviors

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

764

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 most likely enhance memory of targeted behaviors in DBT. other published randomized trial that demonstrated the ef-
Subjects assigned to CTBE were also much more likely fectiveness of any other outpatient treatment for suicidal
than those assigned to DBT to be hospitalized for suicide individuals with BPD. Therefore, there is no “standard” treat-
ideation, despite no apparent differences over time in sui- ment with which to compare DBT. Indeed, with 6 previ-
cide ideation for the 2 conditions. An important DBT strat- ous randomized trials25,28-32,54 (more than any other treat-
egy, at least as a first step, is to encourage outpatient use ment for suicidal behavior or BPD), DBT can be considered
of behavioral skills over inpatient treatment even when sui- the current standard. Comparing DBT with an alterna-
cidal risk is high. Consequently, we cannot determine from tive, however, would provide little, if any, scientific infor-
the present data whether the difference in hospitalization mation. Although knowing that one multicomponent treat-
is due to a lesser tendency of DBT therapists to recom- ment is more effective than another might be of clinical value
mend inpatient treatment or to differences in actual need in deciding which treatment to offer, it does not provide
for hospitalization. There is no evidence that inpatient hos- information about mechanisms of treatment and is of little
pitalization is an effective treatment for suicidality. There- help in developing more effective treatments.
fore, it may be that inpatient admissions for suicide ide- The major limitation of this study is that CTBE subjects
ation were actually iatrogenic rather than therapeutic. These dropped out of the entire study more often than DBT sub-
questions need to be addressed in future studies. jects, despite our best efforts to keep all participants enrolled.
Depression, suicide ideation, and reasons for living im- Careful statistical analyses, however, showed that this dif-
proved significantly in both conditions. The absence of sig- ferential treatment dropout does not explain the differences
nificant differences on these measures is similar to our pre- in outcomes. Nonetheless, future studies must find a way
vious findings25 among severely suicidal patients with BPD to keep more suicidal patients with BPD enrolled. Although
and does not appear to be due to inadequate power. In con- we attempted to control for allegiance and adherence to a
trast, a clinical trial among patients with less severe BPD coherent model by selecting expert therapists and by ask-
found superiority of DBT compared with other treat- ing them to use the treatment that they believed to be most
ments in reducing depression and suicide ideation.28 Both effective, results might be limited by the heterogeneity of
findings fit the 2-stage model of DBT that targets out-of- the treatments used by the community experts compared
control behaviors first (stage 1 target) among more severe with the more homogeneous DBT condition.
patients (stage 1 DBT target) and targets “quiet despera- Additional dismantling studies of DBT are also needed
tion” such as depression (stage 2 DBT target) among less to clarify which components of DBT are essential and to
severe patients who demonstrate behavioral control. what degree fidelity to the DBT manual is needed to achieve
The overall dropout rate in CTBE was high. In addition, results comparable to those found herein and in other stud-
significantly more CTBE subjects were assigned a second ies. This research is under way. Finally, further investiga-
or third therapist (24% of the CTBE group) compared with tion is warranted on community-delivered psycho-
DBT subjects (10% of the DBT group). Because of this re- therapy by experts. Weekly non–behavioral CTBE may have
assignment to new therapists and the greater tendency of much to recommend it as treatment for BPD. Nonethe-
CTBE subjects to see a therapist outside the study, the over- less, findings from this study suggest that DBT may in-
all number of individual sessions was similar between con- deed be the superior intervention.
ditions. Not surprisingly, DBT subjects had more group
therapyhours.However,therewerenosignificantdifferences Submitted for Publication: October 20, 2004; final revision
between conditions in the total amount of therapy received received December 1, 2005; accepted December 8, 2005.
when hours of individual, group, day treatment, and inpa- Author Affiliations: Behavioral Research and Therapy
tient treatment were included. Therefore, the simple amount Clinics and Department of Psychology (Drs Linehan,
of therapeutic contact is not a viable explanation for differ- Brown, Heard, Korslund, and Reynolds, Ms Murray, and
encesinoutcomebetweenDBTandCTBE.Apreviousstudy20 Messrs Tutek and Lindenboim) and Department of Psy-
comparing DBT group “skills training” combined with com- chiatry (Dr Comtois), University of Washington, Se-
munity psychotherapy vs community psychotherapy with- attle; Department of Mathematics, West Chester Univer-
out group training skills found no significant benefit of the sity, West Chester, Pa (Dr Gallop), and Western
group component when added to non-DBT psychotherapy. Psychiatric Institute and Clinic, University of Pitts-
However, we cannot rule out its effects when combined with burgh, Pittsburgh, Pa (Dr Reynolds); British Isles DBT
individual DBT. A component analysis study is under way Training, St Louis, Mo (Dr Heard); and Arizona State Hos-
to tease out the effects of the DBT group component. pital, Phoenix (Mr Tutek).
Subjects assigned to DBT were less likely to continue Correspondence: Marsha M. Linehan, PhD, Behavioral
taking psychotropic medications, including antidepres- Research and Therapy Clinics, Department of Psychol-
sant drugs, during the treatment year. The use of psycho- ogy, University of Washington, Campus Box 351525,
tropic medications in general (and antidepressants specifi- Seattle, WA 98195-1525 (linehan@u.washington.edu).
cally) resumed in the follow-up year, despite no apparent Funding/Support: This study was supported by grants
increase in depression or other dysfunctional behaviors. This MH34486 and MH01593 from the National Institute of
suggests that medication use in this population may be in- Mental Health.
fluenced by psychotherapists’ treatment philosophy as much Previous Presentation: This study was presented in part at
as by symptomatic need for medication. the 36th Annual Convention of the Association for Advance-
An innovative characteristic of this research was the de- ment of Behavior Therapy; November 14, 2002; Reno, Nev.
velopment of the CTBE condition to control for nonspe- Acknowledgment: We thank the therapists and clients
cific treatment factors of expert therapy. We could find no at the Behavioral Research and Therapy Clinics, with-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

765

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015
 out whom this research would not have been possible. borderline personality disorder: a randomized controlled trial. Am J Psychiatry.
1999;156:1563-1569.
We also thank Phillip Abrego, Audra Adelberger, Carol 23. Bateman A, Fonagy P. Treatment of borderline personality disorder with psy-
Barber, Marian Bauman, Susan Bland, Lynn Buell, Steve choanalytically oriented partial hospitalization: an 18-month follow-up. Am J
Clancy, Jim Crowe, Arthur Davis, Romalee Davis, Linda Psychiatry. 2001;158:36-42.
24. Lieb K, Zanarini M, Linehan MM, Bohus M. Seminar section: borderline person-
Dimeff, Brian Grant, Michael Gundle, Anna Harvey, Ruth ality disorder. Lancet. 2004;364:453-461.
Herman-Dunn, Marty Hoiness, Gerald Hover, Charlie 25. Linehan MM, Armstrong HE, Suarez A, Allmon D, Heard H. Cognitive-behavioral
Huffine, Claire Isaacson, Treg Isaacson, Susan Jenkins, treatment of chronically parasuicidal borderline patients. Arch Gen Psychiatry.
1991;48:1060-1064.
Connie Kehrer-Sholl, Kelly Koerner, Glenn Leichman, 26. Linehan MM, Tutek DA, Heard HL, Armstrong HE. Interpersonal outcome of cog-
Eric Levensky, Martha Lyttle, Michael Mandel, Joshua nitive behavioral treatment for chronically suicidal borderline patients. Am J
Psychiatry. 1994;151:1771-1776.
McDavid, Deb McGhee, Evelyn Mercier, Anna Meyer, Es- 27. Hoffman RE. Impact of treatment accessibility on clinical course of parasuicidal
ther Neeser, Susan Radant, Shireen Rizvi, Barbara Sarda- patients [letter]. Arch Gen Psychiatry. 1993;50:157-158.
rov, Jennifer Sayrs, Karen Schmaling, Henry Schmidt, John 28. Koons CR, Robins CJ, Tweed JL, Lynch TR, Gonzalez AM, Morse JQ, Bishop GK,
Butterfield MI, Bastian LA. Efficacy of dialectical behavior therapy in women vet-
Schwab, Stacy Shaw-Welch, Stephen Sholl, Thao Truong, erans with borderline personality disorder. Behav Ther. 2001;32:371-390.
Sandra Walker, and Deborah Wothers for their tireless 29. Linehan MM, Schmidt H III, Dimeff LA, Craft JC, Kanter J, Comtois KA. Dialec-
work. Gratitude is also extended to our colleagues in the tical behavior therapy for patients with borderline personality disorder and
drug-dependence. Am J Addict. 1999;8:279-292.
Department of Psychology at the University of Washing- 30. Linehan MM, Dimeff LA, Reynolds SK, Comtois K, Shaw-Welch S, Heagerty P,
ton and to the Office of Human Subjects for their con- Kivlahan DR. Dialectical behavior therapy versus comprehensive validation plus
tinued support of our work. 12-step for the treatment of opioid dependent women meeting criteria for bor-
derline personality disorder. Drug Alcohol Depend. 2002;67:13-26.
31. Verheul R, van den Bosch LMC, Koeter MWJ, de Ridder MAJ, Stijnen T, van den
REFERENCES Brink W. Dialectical behaviour therapy for women with borderline personality dis-
order: 12-month, randomised clinical trial in the Netherlands. Br J Psychiatry.
2003;182:135-140.
1. American Psychiatric Association. Diagnostic and Statistical Manual of Mental 32. Turner RM. Naturalistic evaluation of dialectical behavioral therapy-oriented treat-
Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994. ment for borderline personality disorder. Cognit Behav Pract. 2000;7:413-419.
2. Fallon P. Traveling through the system: the lived experience of people with bor- 33. American Psychiatric Association. Practice guideline for the treatment of pa-
derline personality disorder in contact with psychiatric services. J Psychiatr Ment tients with borderline personality disorder: American Psychiatric Association Prac-
Health Nurs. 2003;10:393-401. tice Guidelines. Am J Psychiatry. 2001;158(suppl):1-52.
3. Nehls N. Borderline personality disorder: the voice of patients. Res Nurs Health. 34. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measur-
1999;22:285-293. ing depression. Arch Gen Psychiatry. 1961;4:561-571.
4. Soloff PH, Lis JA, Kelly T, Cornelius J, Ulrich R. Risk factors for suicidal behav- 35. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for
ior in borderline personality disorder. Am J Psychiatry. 1994;151:1316-1323. Axis I DSM-IV Disorders, Patient Edition (SCID-I/P). New York: Biometrics Re-
5. Zisook S, Goff A, Sledge P, Schuchter SR. Reported suicidal behavior and cur- search Dept, New York State Psychiatric Institute; 1995.
rent suicidal ideation in a psychiatric outpatient clinic. Ann Clin Psychiatry. 1994; 36. White SJ, Freedman LS. Allocation of patients to treatment groups in a con-
6:27-31. trolled clinical study. Br J Cancer. 1978;37:849-857.
6. Soloff PH, Lynch KG, Kelly TM, Malone KM, Mann JJ. Characteristics of suicide 37. First MB, Spitzer RL, Gibbon M, Williams JBW, Benjamin L. User’s Guide for the
attempts of patients with major depressive episode and borderline personality Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II).
disorder: a comparative study. Am J Psychiatry. 2000;157:601-608. New York: Biometrics Research Dept, New York State Psychiatric Institute; 1996.
7. Grove WM, Tellegen A. Problems in the classification of personality disorders. 38. Loranger AW. International Personality Disorder Examination (IPDE) Manual. White
J Personal Disord. 1991;5:31-41. Plains, NY: Cornell Medical Center; 1995.
8. Stone MH. Long-term outcome in personality disorders. Br J Psychiatry. 1993; 39. Dunn LM. Peabody Picture Vocabulary Test–Revised. Circle Pines, Minn: Ameri-
162:299-313. can Guidance Service; 1981.
9. Frances A, Fyer MR, Clarkin JF. Personality and suicide. Ann N Y Acad Sci. 1986; 40. Linehan MM, Comtois KA, Brown MZ, Heard HL, Wagner AM. Suicide Attempt
487:281-293. Self-Injury Interview (SASII): Development, Reliabililty, and Validity of a Scale
10. Linehan MM, Rizvi SL, Shaw-Welch S, Page B. Psychiatric aspects of suicidal to Assess Suicide Attempts and Intentional Self-Injury. Psychol Assess. In press.
behaviour: personality disorders. In: Hawton K, van Heeringen K, eds. Interna- 41. Linehan MM, Goodstein JL, Nielsen SL, Chiles JA. Reasons for staying alive when
tional Handbook of Suicide and Attempted Suicide. Sussex, England: John Wiley you are thinking of killing yourself: the Reasons for Living Inventory. J Consult
& Sons Inc; 2000:147-178. Clin Psychol. 1983;51:276-286.
11. Geller JL. In again, out again: preliminary evaluation of a state hospital’s worst 42. Hamilton M. A Rating Scale for Depression. J Neurol Neurosurg Psychiatry. 1960;
recidivists. Hosp Community Psychiatry. 1986;37:386-390. 23:56-62.
12. Woogh CM. A cohort through the revolving door. Can J Psychiatry. 1986;31:214- 43. Longford NT. Random Coefficient Models. New York, NY: Oxford University Press
221. Inc; 1993.
13. Bender DS, Dolan RT, Skodol AE, Sanislow CA, Dyck IR, McGlashan TH, Shea 44. Bryk AS, Raudenbush SW. Hierarchical Linear Models: Applications and Data
MT, Zanarini MC, Oldham JM, Gunderson JG. Treatment utilization by patients Analysis Methods. Newbury Park, Calif: Sage Publications; 1992.
with personality disorders. Am J Psychiatry. 2001;158:295-302. 45. Littell RC, Milliken GA, Stroup WW, Wolfinger RD. SAS System for Mixed Models.
14. Zanarini MC, Frankenburg FR, Hennen J, Silk KR. Mental health service utiliza- Cary, NC: SAS Institute Inc; 1996.
tion by borderline personality disorder patients and Axis II comparison subjects 46. Khuri AI, Mathew T, Sinha BK. Statistical Tests for Mixed Linear Models. New
followed prospectively for 6 years. J Clin Psychiatry. 2004;65:28-36. York, NY: John Wiley & Sons Inc; 1998.
15. Skodol AE, Buckley P, Charles E. Is there a characteristic pattern to the treat- 47. Hedeker D, Mermelstein RJ. Analysis of longitudinal substance use outcomes
ment history of clinic outpatients with borderline personality? J Nerv Ment Dis. using ordinal random-effects regression models. Addiction. 2000;95(suppl 4)
1983;171:405-410. :S381-S394.
16. Perry JC, Herman JL, van der Kolk BA, Hoke LA. Psychotherapy and psychological 48. Scott SC, Goldberg MS, Mayo NE. Statistical assessment of ordinal outcomes in
trauma in borderline personality disorder. Psychiatr Ann. 1990;20:33-43. comparative studies. J Clin Epidemiol. 1997;50:45-55.
17. Perry JC, Cooper SH. Psychodynamics, symptoms, and outcome in borderline 49. Wolfinger RD, O’Connell ME. Generalized linear models: a pseudo-likelihood
and antisocial personality disorders and bipolar type II affective disorder. In: approach. J Stat Comput Simul. 1993;48:233-243.
McGlashan TH, ed. The Borderline: Current Empirical Research. Washington, DC: 50. Cox DR, Oakes D. Analysis of Survival Data. London, England: Chapman & Hall
American Psychiatric Press; 1985:21-41. Inc; 1984.
18. Tucker L, Bauer SF, Wagner S, Harlam D, Sher I. Long-term hospital treatment 51. Hedeker D, Gibbons RD. Application of random-effects pattern-mixture models
of borderline patients: a descriptive outcome study. Am J Psychiatry. 1987; for missing data in longitudinal studies. Psychol Methods. 1997;2:78 64.
144:1443-1448. 52. Verbeke G. Linear mixed models for longitudinal data. In: Verbeke G, Molen-
19. Linehan MM, Heard HL, Armstrong HE. Naturalistic follow-up of a behavioral treat- berghs G, eds. Linear Mixed Models for Longitudinal Data: Linear Mixed Mod-
ment for chronically parasuicidal borderline patients. Arch Gen Psychiatry. 1993; els: a SAS Oriented Approach: Lecture Notes in Statistics. Vol 126. New York,
50:971-974. NY: Springer Publishing Co Inc; 1997:108-114.
20. Linehan MM. Cognitive-Behavioral Treatment of Borderline Personality Disorder. 53. Motto JA. Suicide prevention for high-risk persons who refuse treatment. Sui-
New York, NY: Guilford Press; 1993. cide Life Threat Behav. 1976;6:223-230.
21. Linehan MM. Skills Training Manual for Treating Borderline Personality Disorder. 54. van den Bosch LMC, Maarten WJ, Koeter MWJ, Stijen T, Verjeul V, van den Brink
New York, NY: Guilford Press; 1993. K. Sustained efficacy of dialectical behaviour therapy for borderline personality
22. Bateman A, Fonagy P. Effectiveness of partial hospitalization in the treatment of disorder. Behav Res Ther. 2005;43:1231-1241.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 63, JULY 2006 WWW.ARCHGENPSYCHIATRY.COM

766

©2006 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/ by a SUNY Binghamton User on 05/11/2015