Research

JAMA Psychiatry | Original Investigation | META-ANALYSIS

Efficacy of Psychotherapies for Borderline

Personality Disorder
A Systematic Review and Meta-analysis
Ioana A. Cristea, PhD; Claudio Gentili, MD, PhD; Carmen D. Cotet, PhD; Daniela Palomba, MD; Corrado Barbui, MD; Pim Cuijpers, PhD

Editorial page 316

IMPORTANCE Borderline personality disorder (BPD) is a debilitating condition, but several Supplemental content
psychotherapies are considered effective.
CME Quiz at
OBJECTIVE To conduct an updated systematic review and meta-analysis of randomized jamanetworkcme.com
clinical trials to assess the efficacy of psychotherapies for BPD populations.

DATA SOURCES Search terms were combined for borderline personality and randomized trials
in PubMed, PsycINFO, EMBASE, and the Cochrane Central Register of Controlled Trials (from
database inception to November 2015), as well as the reference lists of earlier meta-analyses.

STUDY SELECTION Included were randomized clinical trials of adults with diagnosed BPD
randomized to psychotherapy exclusively or to a control intervention. Study selection
differentiated stand-alone designs (in which an independent psychotherapy was compared
with control interventions) from add-on designs (in which an experimental intervention
added to usual treatment was compared with usual treatment alone).

DATA EXTRACTION AND SYNTHESIS Data extraction coded characteristics of trials, participants,
and interventions and assessed risk of bias using 4 domains of the Cochrane Collaboration Risk
of Bias tool (independent extraction by 2 assessors). Outcomes were pooled using a
random-effects model. Subgroup and meta-regression analyses were conducted.

MAIN OUTCOMES AND MEASURES Standardized mean differences (Hedges g) were calculated
using all outcomes reported in the trials for borderline symptoms, self-harm, suicide, health
service use, and general psychopathology at posttest and follow-up. Differential treatment
retention at posttest was analyzed, reporting odds ratios.

RESULTS Thirty-three trials (2256 participants) were included. For borderline-relevant

outcomes combined (symptoms, self-harm, and suicide) at posttest, the investigated
psychotherapies were moderately more effective than control interventions in stand-alone
designs (g = 0.32; 95% CI, 0.14-0.51) and add-on designs (g = 0.40; 95% CI, 0.15-0.65). Author Affiliations: Department of
Results were similar for other outcomes, including stand-alone designs: self-harm (g = 0.32; Clinical Psychology and
Psychotherapy, Babe-Bolyai
95% CI, 0.09-0.54), suicide (g = 0.44; 95% CI, 0.15-0.74), health service use (g = 0.40; 95% University, Cluj-Napoca, Romania
CI, 0.22-0.58), and general psychopathology (g=0.32; 95% CI, 0.09-0.55), with no (Cristea, Cotet); Department of
differences between design types. There were no significant differences in the odds ratios for General Psychology, University of
Padova, Padova, Italy (Cristea, Gentili,
treatment retention (1.32; 95% CI, 0.87-2.00 for stand-alone designs and 1.01; 95% CI,
Palomba); Meta-Research Innovation
0.55-1.87 for add-on designs). Thirteen trials reported borderline-relevant outcomes at Center at Stanford, Stanford
follow-up (g = 0.45; 95% CI, 0.15-0.75). Dialectical behavior therapy (g=0.34; 95% CI, University, Stanford, California
0.15-0.53) and psychodynamic approaches (g=0.41; 95% CI, 0.12-0.69) were the only types (Cristea); Section of Psychiatry,
Department of Neuroscience,
of psychotherapies more effective than control interventions. Risk of bias was a significant
Biomedicine and Movement
moderator in subgroup and meta-regression analyses (slope = 0.16; 95% CI, 0.29 to Sciences, University of Verona,
0.03; P = .02). Publication bias was persistent, particularly for follow-up. Verona, Italy (Barbui); Department of
Clinical Psychology, Amsterdam
CONCLUSIONS AND RELEVANCE Psychotherapies, most notably dialectical behavior therapy Public Health Research Institute, Vrije
Universiteit, Amsterdam, the
and psychodynamic approaches, are effective for borderline symptoms and related Netherlands (Cuijpers).
problems. Nonetheless, effects are small, inflated by risk of bias and publication bias, and Corresponding Author: Ioana A.
particularly unstable at follow-up. Cristea, PhD, Department of Clinical
Psychology and Psychotherapy,
Babe-Bolyai University, 37 Republicii
JAMA Psychiatry. 2017;74(4):319-328. doi:10.1001/jamapsychiatry.2016.4287 St, 400015, Cluj-Napoca, Romania
Published online March 1, 2017. ([email protected]).

(Reprinted) 319

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Research Original Investigation Efficacy of Psychotherapies for Borderline Personality Disorder

B
orderline personality disorder (BPD) is a debilitating
mental disorder characterized by severe instability in Key Points
affect, identity, interpersonal relationships, and behav-
Question What is the efficacy of psychotherapy for borderline
ioral dysregulation.1 Alongside a vast array of comorbidities, personality disorder?
parasuicide (ie, nonlethal intentional self-harming behav-
Findings In this systematic review and meta-analysis of
iors) and suicide are commonly associated problems. More than
randomized clinical trials, outcomes of psychotherapies (most
75% of patients with BPD are believed to engage in deliberate
notably dialectical behavior therapy and psychodynamic
self-harm.2 Suicide rates are estimated to be between 8% and approaches) significantly improved borderline-relevant outcomes
10%,3,4 almost 50 times higher than in the general population.5 (symptoms, self-harm, and suicide) compared with control
Borderline personality disorder is the most common person- interventions. However, differences dissipated in well-designed
ality disorder in clinical populations,5,6 associated with inten- and implemented trials or if the control group was balanced for
sive use of mental health services7,8 even in the absence of a manualization of treatment or the involvement of the study team
in treatment.
full diagnosis.9 Functional impairment is considerable com-
pared with other personality disorders10 and is enduring in the Meaning Psychotherapies specifically designed for borderline
absence of a change in personality psychopathology.11 personality disorder have significant yet modest benefits over
Several psychotherapy approaches were specifically de- treatment as usual, and future independent and well-conducted
trials are needed to clarify the stability and practical relevance of
veloped for the disorder, most notably dialectical behavior
their effects.
therapy (DBT),12 cognitive behavior therapy (CBT),13 and psy-
chodynamic treatments, such as mentalization-based therapy14
or transference-focused psychotherapy.15 Each approach ap-
peared to be more effective than treatment as usual (TAU) for ister of Controlled Trials) using the search term borderline
BPD-related problems, such as suicidality or parasuicidal personality (both text word and Medical Subject Headings
behavior.16-19 Trials of direct comparisons of treatments for BPD term), with a filter for randomized trials (eMethods in the
reported few differences among them.20,21 However, most trials Supplement). We also checked the reference lists of earlier
demonstrating effectiveness were conducted with the direct meta-analyses.22,23
participation of the treatment developer. Previous meta- Studies were included if they were RCTs in which a psy-
analyses of psychotherapeutic treatments for BPD have been chotherapy was compared with a control condition for adults
scarce and used focused criteria for assessing effectiveness, diagnosed as having BPD. Given the diversity and complexity
avoiding combining treatments. One meta-analysis22 of DBT of therapy orientations, we used an inclusive approach in
for BPD reported moderate effects for borderline-relevant out- delineating the psychotherapy and control conditions. We
comes, suicidality, and self-harm. However, analysis re- used a customary definition of psychotherapy emphasizing
stricted to randomized clinical trials (RCTs) showed reduced verbal communication, structured and purposeful therapist-
effects, with nonsignificance for suicidality and self-harm. An- patient encounters, and the establishment of a therapeutic
other meta-analysis23 of RCTs for BPD reported only moder- relationship.24,25 No constraints were placed on the control
ate effects for the comparison between DBT and TAU. group, which could include (but was not restricted to) TAU or
Conversely, because effectiveness differences between other treatments not specifically developed for BPD. Compari-
therapies appear to be limited and because variations of the sons between 2 different psychotherapies specifically devel-
same intervention are to be expected in diverse implementa- oped for BPD (ie, DBT and transference-focused psycho-
tion settings, we believe that a broader effectiveness evalua- therapy) or between forms of the same psychotherapy (eg, DBT
tion grouping therapies into theoretically intelligible catego- vs its skills training component) were excluded because of ex-
ries is germane. Heterogeneity, publication bias, and potential pectations of similar efficacy. Concomitant medication use was
moderators of efficacy (eg, treatment duration and type of psy- not an exclusion criterion unless it was prescribed in a stan-
chotherapy) are additional unclear issues. Moreover, because dardized way, as in trials in which individuals were random-
the study collection dates of the 2 previous meta-analyses22,23 ized to a combination of psychotherapy and either pharma-
preceded 2012, new trials or follow-ups of older trials pub- cotherapy or placebo. Medication use followed a systematic
lished since then should be considered. Therefore, our objec- protocol, and we could not disentangle its effects from those
tive was to conduct an updated systematic review and meta- of psychotherapy. Studies on even partially adolescent samples
analysis of RCTs to assess the efficacy of psychotherapies were excluded because BPD diagnosis and treatment pose dis-
for BPD-relevant outcomes at posttest and, where possible, at tinct challenges for this group. No language restrictions were
follow-up. applied. One researcher (I.A.C.) screened all records, and full
texts were obtained for RCTs. Two independent assessors
(I.A.C. and C.D.C.) examined the full texts and selected eli-
gible RCTs.
Methods
Identification and Selection of Studies Risk of Bias and Data Extraction
Studies were identified through searches in 4 bibliographical Trial risk of bias (RoB) was evaluated within 4 domains of the
databases (from database inception to November 2015 in Cochrane Collaboration Risk of Bias tool,26 which assesses
PubMed, PsycINFO, EMBASE, and the Cochrane Central Reg- sources of bias in RCTs. Rated domains included (1) adequate

320 JAMA Psychiatry April 2017 Volume 74, Number 4 (Reprinted) jamapsychiatry.com

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Efficacy of Psychotherapies for Borderline Personality Disorder Original Investigation Research

generation of allocation sequence, (2) concealment of alloca- of the reasons. Adverse effects were defined as participant death
tion to conditions, (3) prevention of knowledge of the allo- by suicide and as death from any cause after randomization.
cated intervention to assessors of outcome (masking of asses- The between-group effect size was calculated as the dif-
sors), and (4) dealing with incomplete data. This last domain ference between the intervention and control groups at post-
was assessed as low risk if proper intent-to-treat (ITT) analy- test and at follow-up (Hedges g), corrected for small sample
ses were conducted, meaning that authors used a method for sizes.27 Follow-up data more than 2 years from treatment ter-
imputing missing values so that all randomized participants mination or in which the control group also received the ex-
were included in the analyses. Masking of assessors was rated perimental treatment were not included. Data across mul-
as low risk if the trial described proper methods of ensuring tiple follow-up points were averaged at the study level for each
masking or if all relevant outcome measures were self-report outcome category. For treatment retention, odds ratios indi-
instruments, thus not requiring the direct interaction with an cated the odds of maintaining participants in treatment in the
assessor. For use in meta-regression analyses, we computed intervention group vs the control group.
an overall RoB score for each study by awarding 1 point for each If a trial reported data on multiple outcomes in the same
bias source rated as low risk. category, the mean effect size was calculated using the pro-
Among trials, we distinguished between stand-alone de- cedures by Borenstein et al28 so that each trial reported just
signs (in which the experimental group received a full course one effect size in a category at each time point. Where avail-
of an independent BPD psychotherapy and the control group able, the means (SDs) were used, but if only dichotomous out-
received TAU or another therapy not specific for BPD) and comes were reported, we used available procedures28 to com-
add-on designs (in which both groups received TAU and the pute the standardized mean difference. If a study did not
experimental group received an additional BPD therapy). We include sufficient data for effect size calculation, the authors
also extracted characteristics of the participants, interven- were contacted, and the study was excluded if they failed to
tions, and studies, including therapy type (DBT, psychody- provide data. Where available, ITT data were preferred.
namic, CBT, or other); control group (TAU, supportive therapy, Effect sizes for dichotomous outcomes were computed,
or an ad hoc control, designed as part of the trial); whether the adhering to the ITT principle, by reporting the observed or
control group was manualized (ie, followed a treatment pro- imputed number of patients with an event (eg, self-harm)
tocol or manual); involvement or noninvolvement of the study relative to the total number of patients randomized to that
team in treating the control group; the presence or absence of group.
a treatment developer as an author of the trial report; thera- We used a software program (Comprehensive Meta-
pist supervision for the experimental group (by the treat- Analysis, version 3; Biostat) for computing and pooling effect
ment developer or by others); and low RoB (studies rated as sizes, with a random-effects model for pooling effect sizes. We
low risk for 2 vs 3 RoB domains). We also extracted the fol- calculated the number needed to treat using the formulas by
lowing treatment intensity variables: treatment duration (in Kraemer and Kupfer.29 Heterogeneity was assessed with the
months), treatment exposure (in hours, calculated by multi- I2 statistic: 0% indicates no observed heterogeneity, and higher
plying the session duration for individual or group therapy by values indicate increasing heterogeneity, with 25%, 50%, and
the number of sessions either planned or, if available, 75% defining thresholds for low, moderate, and high. We cal-
attended on average for the experimental and control culated 95% CIs around I2 statistics30 using a noncentral 2-
groups), and difference in treatment exposure between the based approach (heterogi module for Stata, version 8; Stata-
intervention and control groups (in hours). Risk of bias Corp LP).31 Outliers were defined as studies in which the 95%
assessment and data extraction were performed by 2 inde- CI was outside the 95% CI of the pooled studies (on both sides).
pendent assessors. Interrater agreement statistics were For categorical moderators, we conducted subgroup analy-
computed (eMethods in the Supplement), and disagreement ses using a mixed-effects model.28 For continuous modera-
was resolved by discussion among assessors and with the tors, meta-regression analyses used a restricted maximum
senior author (P.C.). likelihood model with the Knapp-Hartung method.28 We ex-
amined publication bias through visual funnel plot inspec-
Meta-analysis tion, the trim-and-fill procedure32 (which produces an effect
To capture the breadth and variability of reported outcomes, size estimate after accounting for publication bias), and
we grouped them into the following categories: borderline rel- Egger test for funnel plot asymmetry.
evant (BPD symptom measures, self-harm and parasuicide, and
suicidal behaviors), borderline symptoms (only BPD symp-
tom measures), self-harm and parasuicidal behavior, suicidal
behavior, health service use (hospitalizations [whether psy-
Results
chiatric or general], use of emergency services, use of addi- Selection and Inclusion of Studies
tional outpatient services, and medication use), and general We screened 1058 abstracts, removed 500 duplicates, and sub-
psychopathology (general psychopathology, anxiety, or de- sequently retrieved 158 full-text articles. Thirty-eight trials ex-
pression). Treatment retention was computed as the compara- amined a psychotherapy, with 5 excluded for comparing 2
tive dropout rates between the intervention and control groups. versions of the same therapy. Consequently, 33 trials (2256
Dropouts (eMethods in the Supplement) were defined as all participants) met our inclusion criteria, and 28 of them
randomized participants not finishing treatment, regardless had enough data for calculating effect sizes (eFigure 1 in the

jamapsychiatry.com (Reprinted) JAMA Psychiatry April 2017 Volume 74, Number 4 321

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Research Original Investigation Efficacy of Psychotherapies for Borderline Personality Disorder

Table 1. Main Effects at Posttest and Follow-up of Trials Comparing Experimental Psychotherapy and Control Treatments
for Borderline Personality Disorder

Stand-alone Design Add-on Design

No. of Hedges g (95% I2 (95% CI), No. of Hedges g (95% I2 (95% CI),
Variable Trials CI)a NNT % Trials CI)a NNT % P Valueb
Posttest
Borderline-relevant outcomesc 17 0.32 5.56 49 10 0.40 4.50 50 .63
(0.14 to 0.51) (0 to 69) (0.15 to 0.65) (0 to 74)
Borderline symptoms 10 0.31 5.75 62 8 0.56 3.25 76 .30
(0.04 to 0.57) (3 to 79) (0.15 to 0.97) (43 to 87)
Self-harm and parasuicidal 13 0.32 5.56 55 6 0.24 7.46 41 .68
behavior (0.09 to 0.54) (0 to 75) (0.07 to 0.55) (0 to 75)
Suicide 10 0.44 4.10 60 3 0.35 5.10 10 .67
(0.15 to 0.74) (0 to 78) (0.02 to 0.68) (0 to 75)
Health service use 13 0.40 4.50 22 3 0.16 11.11 5 .17
(0.22 to 0.58) (0 to 59) (0.13 to 0.46) (0 to 74)
General psychopathology, 13 0.32 5.56 62 10 0.53 3.42 62 .25
anxiety, and depression (0.09 to 0.55) (18 to 78) (0.24 to 0.82) (4 to 79)
Follow-up
Borderline-relevant outcomesc 7 0.56 3.25 62 6 0.35 5.10 79 .51
(0.17 to 0.95) (0 to 81) (0.15 to 0.85) (41 to 89)
Borderline symptoms 3 0.34 5.26 64 4 0.43 4.20 87 .85
(0.13 to 0.81) (0 to 88) (0.41 to 1.26) (62 to 93)
Self-harm and parasuicidal 5 0.58 3.14 74 4 0.04 45.45 0 .07
behavior (0.06 to 1.10) (0 to 88) (0.21 to 0.30) (0 to 68)
Suicide 5 0.34 5.26 39 2 0.31 5.75 0 .90
(0.06 to 0.74) (0 to 76) (0.04 to 0.66)
Health service use 4 0.30 5.95 51 2 0.06 29.41 0 .39
(0.10 to 0.70) (0 to 82) (0.32 to 0.44)
General psychopathology, 5 0.15 11.90 93 5 0.40 4.50 78 .33
anxiety, and depression (1.12 to 0.83) (88 to 96) (0.11 to 0.91) (27 to 89)
Abbreviation: NNT, number needed to treat. group of trials with stand-alone vs add-on designs is significant.
a c
According to the random-effects model. A positive effect indicates superiority Borderline-relevant outcomes include borderline symptoms, self-harm and
of the experimental psychotherapy over control treatments. parasuicidal behavior, and suicide.
b
The P values indicate whether the difference between the effect sizes in the

Supplement). For the 5 missing trials,20,33-36 the authors were location, and 20 trials implemented masking of outcome
contacted, but they did not provide the requested data. assessors (2 used self-report measures only). However, for in-
complete outcome data, only 13 trials were rated as low RoB,
Characteristics of Included Studies and more than half had high RoB. Eleven trials could be rated
The 33 trials included 1169 participants in the investigated treat- as low risk on 3 or 4 domains.
ment group and 1087 participants in the control group (eTable
1 in the Supplement). Seventeen trials targeted patients with Main Effects at Posttest
BPD diagnosed using a structured clinical interview, 10 trials Stand-alone Designs
targeted patients with BPD with recent documented self- Results showed significant effects for all outcome categories,
harm, and the rest targeted special BPD populations (eg, vet- ranging from 0.31 (0.04-0.57) for borderline symptoms to 0.44
erans and individuals with addiction). Twenty-two trials had (0.15-0.74) for suicide outcomes (Table 1). Heterogeneity was
a stand-alone design, and 11 trials had an add-on design. Twelve moderate to high with the exception of health service use. For
trials had women-only samples, and this percentage ranged all borderline-relevant outcomes (eFigure 3 in the Supple-
from 0% to 95% in the remainder. The best-represented ap- ment), 17 trials had a Hedges g of 0.32 (95% CI, 0.14-0.51), with
proaches were DBT (12 trials), psychodynamic therapies (8 moderate heterogeneity (48%).
trials), and CBT (5 trials). Twenty-one trials had TAU as the con-
trol treatment, and the control treatment was manualized in Add-on Designs
10 trials. The treatment developer was an author in 20 trials. For borderline-relevant outcomes (Table 1 and eFigure 3 in the
In 15 trials, the treatment developer was a therapist or super- Supplement), 10 trials had a Hedges g of 0.40 (95% CI, 0.15-
vised therapists directly. Treatment duration ranged from 2.5 0.65), with moderate heterogeneity (50%). Results were not
to 24 months, and the number of sessions (for individual and significant for self-harm and parasuicidal behavior or for health
group therapy taken together) ranged from 6 to 312. service use, but the number of trials was small (range, 3-6).
The statistics indicated high interrater agreement for RoB
estimations (eMethods in the Supplement), which were vari- All Trials
able (eFigure 2 in the Supplement). Sixteen trials reported ad- Combining both design types for all borderline-relevant
equate sequence generation, 12 trials properly concealed al- outcomes (combining borderline symptoms, self-harm and

322 JAMA Psychiatry April 2017 Volume 74, Number 4 (Reprinted) jamapsychiatry.com

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Efficacy of Psychotherapies for Borderline Personality Disorder Original Investigation Research

parasuicidal behavior, and suicide) yielded a significant ef- Meta-regression Analyses

fect (g = 0.35; 95% CI, 0.20-0.50 [27 trials]), with moderate Risk of bias (the number of criteria with low RoB) had a sig-
heterogeneity (48%) (Table 2 and Figure). Results with outli- nificant negative association with outcomes at posttest
ers excluded or excluding comparisons with supportive therapy (slope = 0.16; 95% CI, 0.29 to 0.03; P = .02) (eFigure 5
were similar. in the Supplement). Dropout rates in the treatment group,
For treatment retention, results were not significant for treatment duration, treatment exposure in the treatment
stand-alone designs (odds ratio, 1.32; 95% CI, 0.87-2.00 [15 group or the control group, and difference in treatment
trials]) or add-on designs (odds ratio, 1.01; 95% CI, 0.55-1.87 exposure between groups were not significantly related to
[10 trials]), and heterogeneity was moderate to high (59% outcomes.
for stand-alone designs and 44% for add-on designs). There
were no significant differences between the 2 design types Publication Bias
on any of the symptom outcome categories or on treatment Inspection of the funnel plot and the trim-and-fill procedure
retention. documented publication bias for borderline-relevant out-
comes (eFigure 6 in the Supplement). At posttest, consider-
Main Effects at Follow-up ing all trials, adjustment for missing studies (n = 6) decreased
Adverse Effects the effect size from a Hedges g of 0.35 (95% CI, 0.20-0.50) to
There were 2 deaths by suicide in the treatment group and 5 0.23 (95% CI, 0.07-0.38). For stand-alone designs, 4 studies
deaths by suicide in the control group. The treatment group were imputed, leading to a smaller but significant Hedges g of
and the control group each had 6 all-cause deaths. 0.20 (95% CI, 0.01-0.39). For add-on designs, 2 studies were
imputed, resulting in a Hedges g of 0.30 (95% CI, 0.05-0.56).
Stand-alone Designs Egger test was not significant in any of these cases. At follow-
Seven trials had a significant effect of 0.56 (95% CI, 0.17- up, considering all trials, adjustment for missing studies (n = 3)
0.95) for all borderline-relevant outcomes (Table 1). Hetero- led to a nonsignificant Hedges g of 0.19 (95% CI, 0.15 to 0.53),
geneity was high (62%). The number of trials was too small and Egger test was significant (intercept = 2.78; 95% CI, 0.18-
(range, 3-5) for the other outcome categories. 5.39; P = .04).

Add-on Designs
Six trials had a nonsignificant effect (g = 0.35; 95% CI, 0.15
to 0.85) for all borderline-relevant outcomes (Table 1). Hetero-
Discussion
geneity was high (79%). There were too few trials (range, 2-5) We conducted an updated systematic review and meta-
in the other outcome categories. analysis of RCTs of psychotherapies for BPD. We included 33
trials, either stand-alone designs (an independent experimen-
All Trials tal treatment vs TAU or another control) or add-on designs (an
Combining both design types yielded a significant effect experimental treatment superimposed to TAU vs TAU alone).
(g = 0.45; 95% CI, 0.15-0.75 [13 trials]), with high heteroge- We examined disorder-specific outcomes (BPD symptoms, self-
neity (70%). These results are summarized in Table 2 and in harm and parasuicide, and suicide) and more general out-
eFigure 4 in the Supplement. comes, such as psychopathology or health service use. Most
trials (22 of 33) had stand-alone designs, and our results showed
Subgroup and Meta-regression Analyses significant, small, posttest between-group effect sizes, with
Subgroup Analyses high to moderate heterogeneity across all outcome catego-
These analyses were conducted on the most inclusive out- ries. Results were more variable for add-on designs, includ-
come category (all borderline-relevant outcomes), combin- ing nonsignificant effects. Nonetheless, the number of trials
ing stand-alone and add-on designs because we found no for some outcome categories was small, and these findings
differences among them (Table 2 and eTable 2 in the should be viewed as tentative and possibly spurious. How-
Supplement). The DBT (g=0.34; 95% CI, 0.15-0.53 [9 trials]) ever, for borderline-relevant outcomes and psychopathology
and psychodynamic approaches (g = 0.41; 95% CI, 0.12-0.69 (for which most add-on trials included measures), effects were
[7 trials]) were more effective than control interventions, small to medium. We operated with this design distinction so
while CBT (g = 0.24; 95% CI, 0.01 to 0.49 [5 trials]) and as not to confound more intensive psychotherapeutic treat-
other interventions (g = 0.38; 95% CI, 0.15 to 0.92 [6 ments with others designed to complement usual treatment,
trials]) were not. although subsequent subgroup analysis found no difference
Trials with an ad hoc control group developed as part of between the 2 design types on any of the outcome categories.
the study, trials in which the control intervention was manu- We also found no differences between types of psychothera-
alized, or trials in which the study team was involved in treat- pies. Most trials focused on DBT followed by psychodynamic
ing the control group, as well trials with low RoB for 3 or 4 approaches, and both types generated significant, small
domains, generated nonsignificant between-group effects. Psy- between-group effect sizes, with low heterogeneity for DBT.
chotherapies were more effective than control interventions Surprisingly, CBT was not superior to control conditions. Al-
in trials with more RoB than in those with less RoB (0.48 vs though this result was based on only 5 trials, heterogeneity
0.11, P = .01). was low.

jamapsychiatry.com (Reprinted) JAMA Psychiatry April 2017 Volume 74, Number 4 323

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Research Original Investigation Efficacy of Psychotherapies for Borderline Personality Disorder

Table 2. Main Effects and Results of Subgroup and Meta-regression Analyses at Posttest and Follow-up
of Trials, Combining Both Design Types, for Borderline-Relevant Outcomes
No. of
Variable Trials Hedges g (95% CI)a I2 (95% CI), % NNT P Valueb
Posttest
Borderline-relevant outcomesc 27 0.35 (0.20 to 0.50) 48 (9 to 66) 5.10 NA
Outliers excluded 26 0.38 (0.25 to 0.51) 33 (0 to 57) 4.72 NA
Comparisons with supportive therapy 24 0.35 (0.19 to 0.50) 48 (5 to 67) 5.10 NA
excluded
Outcomes measured on scales 15 0.26 (0.10 to 0.41) 27 (0 to 60) 6.85 NA
created by the treatment developer
Studies conducted by and outcomes 9 0.36 (0.12 to 0.60) 44 (0 to 72) 5.00 NA
measured on scales created by the
treatment developer
Subgroup analysisd
Dialectical behavior therapy 9 0.34 (0.15 to 0.53) 19 (0 to 62) 5.26
Psychodynamic approaches 7 0.41 (0.12 to 0.69) 42 (0 to 74) 4.39
.87
Cognitive behavior therapy 5 0.24 (0.01 to 0.49) 15 (0 to 69) 7.46
Other interventions 6 0.38 (0.15 to 0.92) 79 (41 to 89) 4.72
Control group
Treatment as usual 18 0.40 (0.25 to 0.56) 22 (0 to 57) 4.50
Supportive therapy 3 0.37 (0.36 to 1.09) 62 (0 to 87) 4.85 .49
Ad hoc control group 6 0.17 (0.17 to 0.52) 73 (13 to 86) 10.42
Control group manualized
No 19 0.39 (0.25 to 0.53) 17 (0 to 52) 4.59
.27
Yes 7 0.16 (0.22 to 0.55) 73 (26 to 86) 11.11
Study team treating the control
group
No 17 0.42 (0.28 to 0.56) 6 (0 to 48) 4.27
.14
Yes 10 0.18 (0.11 to 0.46) 67 (21 to 81) 9.80
Treatment developer a trial author
No 12 0.31 (0.16 to 0.46) 5 (0 to 52) 5.75
.79
Yes 15 0.35 (0.10 to 0.59) 63 (26 to 78) 5.10
Therapist supervision
Treatment developer 11 0.37 (0.13 to 0.62) 52 (0 to 74) 4.85
.49
Other 8 0.26 (0.08 to 0.45) 9 (0 to 60) 6.85
Low risk of bias criteria
0-2 18 0.48 (0.33 to 0.64) 15 (0 to 52) 3.76
.01
3-4 9 0.11 (0.12 to 0.35) 57 (0 to 78) 16.13
Follow-up
Borderline-relevant outcomesc 13 0.45 (0.15 to 0.75) 70 (41 to 82) 4.00 NA
Outliers excluded 12 0.32 (0.08 to 0.55) 52 (0 to 73) 5.56 NA
Comparisons with supportive therapy 12 0.47 (0.16 to 0.78) 73 (45 to 83) 3.85 NA
excluded
Subgroup analysisd
Dialectical behavior therapy 4 0.42 (0.02 to 0.87) 62 (0 to 85) 4.27
Psychodynamic approaches 2 0.40 (1.10 to 1.89) 88 (Not 4.50
available)e
.18
Cognitive behavior therapy 5 0.12 (0.12 to 0.35) 0 (0 to 64) 14.71
Other interventions 2 1.53 (0.11 to 2.99) 84 (Not 1.39
available)e
Control group
Treatment as usual 9 0.52 (0.10 to 0.94) 76 (48 to 86) 3.50
.63
Ad hoc control group 3 0.36 (0.11 to 0.84) 66 (0 to 88) 5.00
Control group manualized
No 10 0.50 (0.13 to 0.87) 73 (42 to 84) 3.62
.61
Yes 3 0.31 (0.31 to 0.93) 64 (0 to 88) 5.75

(continued)

324 JAMA Psychiatry April 2017 Volume 74, Number 4 (Reprinted) jamapsychiatry.com

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Efficacy of Psychotherapies for Borderline Personality Disorder Original Investigation Research

Table 2. Main Effects and Results of Subgroup and Meta-regression Analyses at Posttest and Follow-up
of Trials, Combining Both Design Types, for Borderline-Relevant Outcomes (continued)
Abbreviations: NNT, number needed
No. of
Variable Trials Hedges g (95% CI)a I2 (95% CI), % NNT P Valueb to treat; NA, not applicable.
a
Study team treating the control According to the random-effects
group model.
b
No 8 0.63 (0.19 to 1.07) 76 (43 to 87) 2.91 The P values indicate whether the
.15 difference between the effect sizes
Yes 5 0.20 (0.19 to 0.58) 52 (0 to 81) 8.93
in the subgroups is significant.
Treatment developer a trial author c
Borderline-relevant outcomes
No 4 0.14 (0.09 to 0.37) 0 (0 to 68) 12.82 include borderline symptoms,
.09
Yes 9 0.57 (0.12 to 1.02) 77 (51 to 87) 3.18 self-harm and parasuicidal behavior,
Therapist supervision and suicide.
d
Subgroup analyses were conducted
Treatment developer 7 0.32 (0.07 to 0.73) 63 (0 to 82) 5.56
.49 using a mixed-effects model. Only
Other 4 0.62 (0.11 to 1.36) 85 (54 to 93) 2.96 subgroups with at least 2 trials were
Low risk of bias criteria included.
e
0-2 9 0.56 (0.13 to 0.99) 75 (45 to 86) 3.25 The 95% CIs around I2 cannot be
.24 calculated if there are fewer than 3
3-4 4 0.23 (0.12 to 0.57) 46 (0 to 81) 7.69
subgroups.

Figure. Borderline-Relevant Outcomes

Favors Favors
Source Hedges g (95% CI) Control Psychotherapy
Amianto et al,43 2011 0.34 (1.01 to 0.34)
Bateman and Fonagy,37 1999 1.06 (0.28 to 1.83)
Bateman and Fonagy,16 2009 0.75 (0.28 to 1.23)
Blum et al,44 2008 0.31 (0.04 to 0.66)
Bos et al,41 2010 0.22 (0.30 to 0.74)
Carter et al,45 2010 0.10 (0.45 to 0.66)
Cottraux et al,46 2009 0.48 (1.53 to 0.56)
Davidson et al,17 2006 0.10 (0.30 to 0.51)
Doering et al,18 2010 0.25 (0.13 to 0.64)
Farrell et al,47 2009 1.04 (0.25 to 1.84)
Gratz and Gunderson,48 2006 1.02 (0.16 to 1.88)
Gratz et al,49 2014 0.89 (0.35 to 1.43)
Gregory et al,50 2008 0.26 (0.51 to 1.03)
Jrgensen et al,51 2013 0.35 (0.15 to 0.85)
Koons et al,42 2001 0.62 (0.25 to 1.50)
Kramer et al,52 2014 0.07 (0.38 to 0.52)
Leppnen et al,53 2016 0.25 (0.31 to 0.81)
Linehan et al,38 1991 0.49 (0.14 to 1.11)
Linehan et al,19 2006 0.30 (0.12 to 0.72)
McMain et al,40 2009 0.02 (0.27 to 0.31)
Pascual et al,54 2015 0.73 (1.32 to 0.14)
Priebe et al,55 2012 0.42 (0.05 to 0.89)
Reneses et al,56 2013 0.55 (0.05 to 1.16)
Soler et al,57 2009 0.56 (0.04 to 1.08)
Turner,58 2000 1.07 (0.24 to 1.91) Shown are standardized posttest
Verheul et al,39 2003 0.51 (0.14 to 1.16) effect sizes of comparisons between
Weinberg et al,59 2006 0.82 (0.06 to 1.57) investigated psychotherapies and
Overall 0.35 (0.20 to 0.50) control conditions for
borderline-relevant outcomes
2.0 1.0 0 1.0 2.0 (borderline symptoms, self-harm and
Hedges g (95% CI) parasuicidal behavior, and suicide) for
27 trials.16-19,37-59

Follow-up results for all borderline-relevant outcomes increases the risk of biasing effects from factors extraneous to
showed significant medium effects for stand-alone designs and the intervention. Adverse effects were rare in both the experi-
nonsignificant results for add-on designs. Heterogeneity was mental and control treatment groups.
high, reflecting substantial differences in follow-up duration A surprising finding regarded treatment retention, for
and type among trials, with some being naturalistic and oth- which we found no significant differences between the ex-
ers including booster interventions, which in some cases were perimental treatment and control groups. This result re-
intensive.37 We did not include follow-up points more than 2 mained stable for both design types and showed moderate
years after treatment termination because a longer time span heterogeneity. Improving treatment retention has generally

jamapsychiatry.com (Reprinted) JAMA Psychiatry April 2017 Volume 74, Number 4 325

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Research Original Investigation Efficacy of Psychotherapies for Borderline Personality Disorder

been seen as a substantial advantage of psychotherapies for comes. Moreover, there was a linear relationship between the
BPD, and most trials in general have demonstrated very favor- number of criteria with low RoB and effect sizes: effects de-
able evaluations. This discrepancy might stem from the fact creased by 0.16 for each additional domain that could be rated
that individual trials used variable ways of calculating drop- as low RoB. There was also evidence of publication bias for
out rates, while we used a standard ITT method whereby all posttest results and particularly for follow-up. Its potential
participants who did not finish treatment after randomiza- adjustment reduced effect sizes to smaller, albeit significant,
tion were considered dropouts regardless of whether they values for both design types at posttest but to nonsignificant
started treatment or what their specific reasons were for dis- values for both design types at follow-up.
continuing it. Moreover, in calculating rates, we reported the
absolute number of dropouts relative to the number of ran- Limitations
domized participants in each group. In contrast, individual Our meta-analysis has several limitations. Some outcome cat-
trials used very diverse methods for defining dropouts, such egories or subgroups included data from a small number of
as considering participants who switched therapists19,38,39 or trials, rendering resultant effect sizes potentially uncertain. For
who missed 4 consecutive sessions.40 In other cases, partici- 5 trials, we did not obtain access to data necessary for calcu-
pants who did not initiate treatment were not counted when lating effect sizes. Furthermore, most trials had not been reg-
calculating dropout rates.39,41,42 One meta-analysis60 of treat- istered in clinical trial registries, so we could not rate RoB
ment completion in BPD reported high treatment completion because of selective outcome reporting. Our search was broad,
rates (approximately 75%) but with substantial between- but we may have missed trials that addressed personality dis-
study heterogeneity. However, included studies were both ob- orders in general but ultimately had a sample composed of pa-
servational and controlled trials, randomized or not. Unlike tients with BPD. Owing to the small number of trials, we
herein, the authors did not calculate differences in treatment grouped therapies in broader categories, effacing subtler dif-
completion between the experimental and control groups. In- ferences between orientations. Frequently cited approaches,
stead, they pooled absolute rates for the former, a procedure such as schema-focused therapy, were underrepresented,
that is discouraged because it can lead to extremely high lev- mainly because they were mostly studied in head-to-head
els of heterogeneity.61 trials. The use of adjunct medication was neither standard-
We further investigated the potential sources of hetero- ized nor consistently reported and could have confounded
geneity in subgroup and meta-regression analyses. More than psychotherapy effects.
half of the trials included the treatment developer as an au-
thor, but they generated similar effects as independent trials.
Yet a subtler effect potentially related to the involvement of
the treatment developer emerged: differences between the ex-
Conclusions
perimental treatment and control groups were no longer sig- Various independent psychotherapies demonstrated effi-
nificant in trials with an ad hoc control group developed as part cacy for borderline-relevant symptoms, self-harm, suicide,
of the study, where the control intervention was manualized, health service use, and general psychopathology in BPD.
or where the study team was involved in treating the control However, effects were small, inflated by publication bias, and
group. We can speculate that, at least in part, the differential particularly unstable for follow-up. These effects were no lon-
efficacy of psychotherapies designed for BPD in contrast to ger sustained in trials with low RoB. While treatment inten-
usual treatment could be due to the special attention granted sity per se did not seem to influence outcomes, there are
to the experimental group or indeed to having a manualized, indications that a control group balanced for the involve-
structured treatment. Nevertheless, treatment intensity (both ment of the study team in treatment or with a manualized
treatment duration and exposure) was not related to the treat- protocol is as effective as psychotherapies tailored for BPD.
ment outcomes considered. We found no evidence that treatment retention would be
Trial RoB consistently emerged as a moderator of effect higher for specific psychotherapies than for control inter-
sizes in both subgroup and meta-regression analyses. Trials ventions, contradicting systematic claims from individual
with low RoB for at least 3 of the 4 domains considered gen- trials. Future trials should implement prospective registra-
erated nonsignificant effects for borderline-relevant out- tion in clinical trial registries.

ARTICLE INFORMATION Critical revision of the manuscript for important Scientific Research and Innovation, Consiliului
Accepted for Publication: December 24, 2016. intellectual content: Gentili, Cotet, Palomba, Barbui, Naional al Cercetrii tiinificeUnitatea Executiv
Cuijpers. Pentru Finanarea nvmntului Superior, a
Published Online: March 1, 2017. Statistical analysis: Cristea, Barbui, Cuijpers. Cercetrii, Dezvoltrii i Inovrii (awarded to Dr
doi:10.1001/jamapsychiatry.2016.4287 Obtained funding: Cristea. Cristea). Dr Cristea was also supported by a Visiting
Author Contributions: Dr Cristea had full access to Administrative, technical, or material support: Scientist Grant from the University of Padova.
all the data in the study and takes responsibility for Gentili, Barbui. Role of the Funder/Sponsor: Neither funding
the integrity of the data and the accuracy of the Study supervision: Cristea, Barbui, Cuijpers. organization had any role in the design and conduct
data analysis. Conflict of Interest Disclosures: None reported. of the study; collection, management, analysis, and
Study concept and design: Cristea, Gentili, Cuijpers. interpretation of the data; preparation, review, or
Acquisition, analysis, or interpretation of data: Funding/Support: Drs Cristea and Cotet were
supported for this work by grant PN-II-RU-TE-2014 approval of the manuscript; and decision to submit
Cotet, Palomba, Barbui, Cuijpers. the manuscript for publication.
Drafting of the manuscript: Cristea, Gentili. -4-1316 from the Romanian National Authority for

326 JAMA Psychiatry April 2017 Volume 74, Number 4 (Reprinted) jamapsychiatry.com

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Efficacy of Psychotherapies for Borderline Personality Disorder Original Investigation Research

REFERENCES 18. Doering S, Hrz S, Rentrop M, et al. on symptoms and quality of marital relationships
1. American Psychiatric Association. Diagnostic and Transference-focused psychotherapy v. treatment and mental health of Iranian borderline personality
Statistical Manual of Mental Disorders. 5th ed. by community psychotherapists for borderline couples: a controlled trial. Interdiscip J Contemp Res
Arlington, VA: American Psychiatric Association; 2013. personality disorder: randomised controlled trial. Br Bus. 2012;3(9):1480-1487.
J Psychiatry. 2010;196(5):389-395. 34. Linehan MM, Schmidt H III, Dimeff LA, Craft JC,
2. Oldham JM. Borderline personality disorder and
suicidality. Am J Psychiatry. 2006;163(1):20-26. 19. Linehan MM, Comtois KA, Murray AM, et al. Kanter J, Comtois KA. Dialectical behavior therapy
Two-year randomized controlled trial and follow-up for patients with borderline personality disorder
3. American Psychiatric Association Practice of dialectical behavior therapy vs therapy by and drug-dependence. Am J Addict. 1999;8(4):279-
Guidelines. Practice guideline for the treatment of experts for suicidal behaviors and borderline 292.
patients with borderline personality disorder. Am J personality disorder. Arch Gen Psychiatry. 2006;63
Psychiatry. 2001;158(10)(suppl):1-52. 35. Linehan MM, Dimeff LA, Reynolds SK, et al.
(7):757-766. Dialectical behavior therapy versus comprehensive
4. Black DW, Blum N, Pfohl B, Hale N. Suicidal 20. Clarkin JF, Levy KN, Lenzenweger MF, validation therapy plus 12-step for the treatment of
behavior in borderline personality disorder: Kernberg OF. Evaluating three treatments for opioid dependent women meeting criteria for
prevalence, risk factors, prediction, and prevention. borderline personality disorder: a multiwave study. borderline personality disorder. Drug Alcohol Depend.
J Pers Disord. 2004;18(3):226-239. Am J Psychiatry. 2007;164(6):922-928. 2002;67(1):13-26.
5. Leichsenring F, Leibing E, Kruse J, New AS, 21. Giesen-Bloo J, van Dyck R, Spinhoven P, et al. 36. Munroe-Blum H, Marziali E. A controlled trial of
Leweke F. Borderline personality disorder. Lancet. Outpatient psychotherapy for borderline short-term group treatment for borderline
2011;377(9759):74-84. personality disorder: randomized trial of personality disorder. J Pers Disord. 1995;9(3):190-
6. Gunderson JG. Borderline personality disorder: schema-focused therapy vs transference-focused 198. doi:10.1521/pedi.1995.9.3.190
ontogeny of a diagnosis. Am J Psychiatry. 2009;166 psychotherapy. Arch Gen Psychiatry. 2006;63(6): 37. Bateman A, Fonagy P. Effectiveness of partial
(5):530-539. 649-658. hospitalization in the treatment of borderline
7. Zanarini MC, Frankenburg FR, Hennen J, Silk KR. 22. Kliem S, Krger C, Kosfelder J. Dialectical personality disorder: a randomized controlled trial.
Mental health service utilization by borderline behavior therapy for borderline personality Am J Psychiatry. 1999;156(10):1563-1569.
personality disorder patients and Axis II comparison disorder: a meta-analysis using mixed-effects 38. Linehan MM, Armstrong HE, Suarez A, Allmon
subjects followed prospectively for 6 years. J Clin modeling. J Consult Clin Psychol. 2010;78(6):936- D, Heard HL. Cognitive-behavioral treatment of
Psychiatry. 2004;65(1):28-36. 951. chronically parasuicidal borderline patients. Arch
8. Zanarini MC, Frankenburg FR, Khera GS, 23. Stoffers JM, Vllm BA, Rcker G, Timmer A, Gen Psychiatry. 1991;48(12):1060-1064.
Bleichmar J. Treatment histories of borderline Huband N, Lieb K. Psychological therapies for 39. Verheul R, Van Den Bosch LM, Koeter MW, De
inpatients. Compr Psychiatry. 2001;42(2):144-150. people with borderline personality disorder. Ridder MA, Stijnen T, Van Den Brink W. Dialectical
9. Ten Have M, Verheul R, Kaasenbrood A, et al. Cochrane Database Syst Rev. 2012;(8):CD005652. behaviour therapy for women with borderline
Prevalence rates of borderline personality disorder 24. Cox GR, Callahan P, Churchill R, et al. personality disorder: 12-month, randomised clinical
symptoms: a study based on the Netherlands Psychological therapies versus antidepressant trial in the Netherlands. Br J Psychiatry. 2003;182:
Mental Health Survey and Incidence Study-2. BMC medication, alone and in combination for 135-140.
Psychiatry. 2016;16(1):249. depression in children and adolescents. Cochrane 40. McMain SF, Links PS, Gnam WH, et al.
10. Skodol AE, Gunderson JG, McGlashan TH, et al. Database Syst Rev. 2014;(11):CD008324. A randomized trial of dialectical behavior therapy
Functional impairment in patients with schizotypal, 25. Barth J, Munder T, Gerger H, et al. Comparative versus general psychiatric management for
borderline, avoidant, or obsessive-compulsive efficacy of seven psychotherapeutic interventions borderline personality disorder. Am J Psychiatry.
personality disorder. Am J Psychiatry. 2002;159(2): for patients with depression: a network 2009;166(12):1365-1374.
276-283. meta-analysis. PLoS Med. 2013;10(5):e1001454. 41. Bos EH, van Wel EB, Appelo MT, Verbraak MJ.
11. Skodol AE, Pagano ME, Bender DS, et al. doi:10.1371/journal.pmed.1001454 A randomized controlled trial of a Dutch version of
Stability of functional impairment in patients with 26. Higgins JP, Altman DG, Gtzsche PC, et al; systems training for emotional predictability and
schizotypal, borderline, avoidant, or Cochrane Bias Methods Group; Cochrane Statistical problem solving for borderline personality disorder.
obsessive-compulsive personality disorder over Methods Group. The Cochrane Collaborations tool J Nerv Ment Dis. 2010;198(4):299-304.
two years. Psychol Med. 2005;35(3):443-451. for assessing risk of bias in randomised trials. BMJ. 42. Koons CR, Robins CJ, Tweed JL, et al. Efficacy
12. Linehan MM. Cognitive-Behavioral Treatment of 2011;343:d5928. of dialectical behavior therapy in women veterans
Borderline Personality Disorder. New York, NY: 27. Hedges LV, Olkin I. Statistical Methods for with borderline personality disorder. Behav Ther.
Guilford Press; 1993. Meta-Analysis. Orlando, FL: Academic Press; 1985. 2001;32(2):371-390. doi:10.1016/S0005-7894(01)
13. Beck AT, Freeman A. Cognitive Therapy of 28. Borenstein M, Hedges LV, Higgins JPT, 80009-5
Personality Disorders. New York, NY: Guilford Press; Rothstein HR. Introduction to Meta-Analysis. 43. Amianto F, Ferrero A, Pier A, et al. Supervised
1990. Chichester, England: Wiley; 2009. team management, with or without structured
14. Bateman A, Fonagy P. Psychotherapy for 29. Kraemer HC, Kupfer DJ. Size of treatment psychotherapy, in heavy users of a mental health
Borderline Personality Disorder: Mentalisation effects and their importance to clinical research and service with borderline personality disorder:
Based Treatment. Oxford, England: Oxford University practice. Biol Psychiatry. 2006;59(11):990-996. a two-year follow-up preliminary randomized study.
Press; 2004. BMC Psychiatry. 2011;11:181.
30. Ioannidis JP, Patsopoulos NA, Evangelou E.
15. Clarkin JF, Yeomans FE, Kernberg OE. Uncertainty in heterogeneity estimates in 44. Blum N, St John D, Pfohl B, et al. Systems
Psychotherapy for Borderline Personality. New York, meta-analyses. BMJ. 2007;335(7626):914-916. Training for Emotional Predictability and Problem
NY: John Wiley & Sons; 1999. Solving (STEPPS) for outpatients with borderline
31. Orsini N, Bottai M, Higgins J, Buchan I. Heterogi: personality disorder: a randomized controlled trial
16. Bateman A, Fonagy P. Randomized controlled Stata module to quantify heterogeneity in a and 1-year follow-up. Am J Psychiatry. 2008;165(4):
trial of outpatient mentalization-based treatment meta-analysis. http://econpapers.repec.org 468-478.
versus structured clinical management for /software/bocbocode/s449201.htm. Published
borderline personality disorder. Am J Psychiatry. 2006. Accessed November 27, 2013. 45. Carter GL, Willcox CH, Lewin TJ, Conrad AM,
2009;166(12):1355-1364. Bendit N. Hunter DBT project: randomized
32. Duval S, Tweedie R. Trim and fill: a simple controlled trial of dialectical behaviour therapy in
17. Davidson K, Norrie J, Tyrer P, et al. The funnel-plotbased method of testing and adjusting women with borderline personality disorder.
effectiveness of cognitive behavior therapy for for publication bias in meta-analysis. Biometrics. Aust N Z J Psychiatry. 2010;44(2):162-173.
borderline personality disorder: results from the 2000;56(2):455-463.
Borderline Personality Disorder Study of Cognitive 46. Cottraux J, Note ID, Boutitie F, et al. Cognitive
33. Kamalabadi MJ, Ahmadi SA, Etemadi O, therapy versus Rogerian supportive therapy in
Therapy (BOSCOT) trial. J Pers Disord. 2006;20(5): Fatehizadeh M, Bahrami F, Firoozabadi A. A study of
450-465. borderline personality disorder: two-year follow-up
the effect of couple dialectical behavioral therapy

jamapsychiatry.com (Reprinted) JAMA Psychiatry April 2017 Volume 74, Number 4 327

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017

 Research Original Investigation Efficacy of Psychotherapies for Borderline Personality Disorder

of a controlled pilot study. Psychother Psychosom. in patients with borderline personality disorder: results of a randomized and controlled trial. Actas
2009;78(5):307-316. a randomized trial. Acta Psychiatr Scand. 2013;127 Esp Psiquiatr. 2013;41(3):139-148.
47. Farrell JM, Shaw IA, Webber MA. (4):305-317. 57. Soler J, Pascual JC, Tiana T, et al. Dialectical
A schema-focused approach to group 52. Kramer U, Kolly S, Berthoud L, et al. Effects of behaviour therapy skills training compared to
psychotherapy for outpatients with borderline motive-oriented therapeutic relationship in a standard group therapy in borderline personality
personality disorder: a randomized controlled trial. ten-session general psychiatric treatment of disorder: a 3-month randomised controlled clinical
J Behav Ther Exp Psychiatry. 2009;40(2):317-328. borderline personality disorder: a randomized trial. Behav Res Ther. 2009;47(5):353-358.
48. Gratz KL, Gunderson JG. Preliminary data on controlled trial. Psychother Psychosom. 2014;83(3): 58. Turner RM. Naturalistic evaluation of dialectical
an acceptance-based emotion regulation group 176-186. behavior therapyoriented treatment for borderline
intervention for deliberate self-harm among 53. Leppnen V, Hakko H, Sintonen H, Lindeman S. personality disorder. Cogn Behav Pract. 2000;7(4):
women with borderline personality disorder. Behav Comparing effectiveness of treatments for 413-419.
Ther. 2006;37(1):25-35. borderline personality disorder in communal 59. Weinberg I, Gunderson JG, Hennen J, Cutter CJ
49. Gratz KL, Tull MT, Levy R. Randomized mental health care: the Oulu BPD Study. Jr. Manual assisted cognitive treatment for
controlled trial and uncontrolled 9-month follow-up Community Ment Health J. 2016;52(2):216-227. deliberate self-harm in borderline personality
of an adjunctive emotion regulation group therapy 54. Pascual JC, Palomares N, Ibez , et al. disorder patients. J Pers Disord. 2006;20(5):482-
for deliberate self-harm among women with Efficacy of cognitive rehabilitation on psychosocial 492.
borderline personality disorder. Psychol Med. 2014; functioning in borderline personality disorder: 60. Barnicot K, Katsakou C, Marougka S, Priebe S.
44(10):2099-2112. a randomized controlled trial. BMC Psychiatry. Treatment completion in psychotherapy for
50. Gregory RJ, Chlebowski S, Kang D, et al. 2015;15:255. borderline personality disorder: a systematic review
A controlled trial of psychodynamic psychotherapy 55. Priebe S, Bhatti N, Barnicot K, et al. and meta-analysis. Acta Psychiatr Scand. 2011;123
for co-occurring borderline personality disorder and Effectiveness and cost-effectiveness of dialectical (5):327-338.
alcohol use disorder. Psychotherapy (Chic). 2008; behaviour therapy for self-harming patients with 61. Higgins JPT, Green S, eds. Cochrane Handbook
45(1):28-41. personality disorder: a pragmatic randomised for Systematic Reviews of Interventions. Version
51. Jrgensen CR, Freund C, Bye R, Jordet H, controlled trial. Psychother Psychosom. 2012;81(6): 5.1.0. http://training.cochrane.org/handbook.
Andersen D, Kjlbye M. Outcome of 356-365. Updated March 2011. Accessed November 2016.
mentalization-based and supportive psychotherapy 56. Reneses B, Galin M, Serrano R, et al. A new
time limited psychotherapy for BPD: preliminary

328 JAMA Psychiatry April 2017 Volume 74, Number 4 (Reprinted) jamapsychiatry.com

Copyright 2017 American Medical Association. All rights reserved.

Downloaded From: http://archpsyc.jamanetwork.com/pdfaccess.ashx?url=/data/journals/psych/936149/ by a University of California - San Diego User on 04/30/2017