Abstract Personality disorders are widely prevalent among those seeking mental health services, resulting in substantial distress and a heavy burden on public assistance and health resources
Correspondence: Alexander Chapman, PhD, Department of Psychology, Simon Fraser University, 8888 University Drive, Burnaby, BC, V5A 4Z1 Canada. Tel: (778) 782-6932. Fax: (778) 782-3427. E-mail: alchapma@sfu.ca
(Received 7 April 2011 ; accepted 5 May 2011 )
Psychotherapy for personality disorders
KATHERINE L. DIXON-GORDON , BRIANNA J. TURNER & ALEXANDER L. CHAPMAN
Department of Psychology, Simon Fraser University, Burnaby, British Columbia, Canada
Abstract Personality disorders are widely prevalent among those seeking mental health services, resulting in substantial distress and a heavy burden on public assistance and health resources. We conducted a qualitative review of randomized controlled trials (RCTs) of psychosocial interventions for personality disorders. Articles were identifi ed through searches of electronic databases and classifi ed based on the focus of the psychological intervention. Data regarding treatment, participants and outcomes were identifi ed. We identifi ed 33 RCTs that evaluated the effi cacy of various psychosocial treatments. Of these studies, 19 focused on treatment of borderline personality disorder, and suggested that there are several effi cacious treat- ments and one well-established treatment for this disorder. In contrast, only fi ve RCTs examined the effi cacy of treatments for Cluster C personality disorders, and no RCTs tested the effi cacy of treatments for Cluster A personality disorders. Although other personality disorders, especially Cluster A, place heavy demands on public assistance, and in spite of recommendations that psychosocial interventions should be the fi rst line of treatment for these disorders, our review underscored the dearth of treatment research for many of these personality disorders. We highlight some obstacles to such research and suggest directions for future research.
Introduction
According to the DSM-IV, personality disorders (PDs) are defi ned as pervasive, non-normative patterns of thought and behaviour which are long- standing, and cause signifi cant impairment in rela- tionships and overall functioning (APA, 2000, p. 685). The DSM-IV includes ten PDs, organized into three clusters: Cluster A disorders, comprising schizoid, paranoid and schizotypal PDs, are charac- terized by odd or eccentric patterns of behaviour; Cluster B disorders, comprising antisocial, border- line, narcissistic and histrionic PDs, are character- ized by dramatic or impulsive patterns of behaviour; and Cluster C disorders, comprising avoidant, dependent, and obsessive – compulsive PDs, are char- acterized by anxious or fearful behaviours. PDs are highly prevalent, with 31 – 45% of psychiatric patients and 10 – 15% of the general adult population meeting criteria for at least one PD (Samuels et al., 2002; Zimmerman & Coryell, 1989).
Personality disorders are associated with substan- tial personal and interpersonal distress, functional impairment, and use of mental health resources (Perry, 1993; Perry & Vaillant, 1989; Skodol, Johnson, Cohen, Sneed, & Crawford, 2007). In fact,
individuals with PDs make up a substantial portion of mental health service consumers (Fyer, Frances, Sullivan, Hurt, & Clarkin, 1988; Markowitz, Moran, Kocsis, & Frances, 1992; Oldham, Skodol, Kellman, & Hyler, 1995; Skodol et al., 1993; Vaughn et al., 2010). Further, early literature documenting limited gains in psychotherapy among individuals with PDs compared to those without PDs (Diguer, Barber, & Luborsky, 1993; Fahy, Eisler, & Russell, 1993; Hardy et al., 1995; Karterud et al., 1992; Shea, Pilkonis, Beckham, & Collins, 1990; Woody, McLellan, Luborsky, & O ’ Brien, 1985) fuelled the assumption that individuals with personality disor- ders may be ‘ untreatable ’ (Lewis & Appleby, 1988).
Although people who hold this assumption have been taken to task with the emergence of mounting evidence for the effi cacy of treatments for PDs, PDs are associated with signifi cant challenges for psycho- therapy. For example, interpersonal, self and identity dysfunction are often hallmark features of various PDs (Livesley, 2003); thus, it is not surprising that the formation and maintenance of a positive working alliance can be a challenging endeavour (Benjamin & Karpiak, 2002; Colson et al., 1985; Muran, Segal, Samstag & Crawford, 1994). Individuals with PDs
International Review of Psychiatry, June 2011; 23: 282–302
ISSN 0954–0261 print/ISSN 1369–1627 online © 2011 Institute of Psychiatry DOI: 10.3109/09540261.2011.586992
Psychotherapy for personality disorders 283
often present to therapy with a variety of challenging behaviours that require attention, including sub- stance use, eating disorders, self-injury, suicidality, and violent or aggressive behaviour (Grant et al., 2004). Moreover, patients with particular PDs pres- ent to treatment with an average of roughly three co-occurring Axis-I disorders (Harned et al., 2009; McMain et al., 2009), making it diffi cult to defi ne and prioritize treatment targets, and to ascertain meaningful ‘ progress ’ . The clinical complexity of these patients can lead to distress, demoralization and burn-out on the part of therapist (Chapman, 2009; Rossberg, Karterud, Pedersen, & Friis, 2008). Individuals with PDs are also more likely to prema- turely terminate therapy compared to those without PDs (Karterud et al., 1992; Skodol, Buckley, & Charles, 1983). Thus, it is not surprising that work with these individuals is sometimes marked by frus- tration on the part of both therapist and patient regarding the rate of therapeutic progress (Murphy & McVey, 2010; Watts & Morgan, 1994).
Research indicates that many mental health pro- fessionals hold a variety of negative beliefs about individuals with PDs, including, for example, that these patients are challenging, attention-seeking, manipulative and even ‘ annoying ’ (Cleary, Siegfried, & Walter, 2002; Fraser & Gallop, 1993; Gallop, Lancee, & Garfi nkel, 1989; James & Cowman, 2007; Lewis & Appleby, 1988). Further, many mental health professionals believe that patients with PDs are less likely to respond to intervention, more likely to pose challenges for clinical management, and are less deserving of mental health resources than indi- viduals without personality disorders (Lewis & Appleby, 1988). Such negative beliefs about indi- viduals with PDs are associated with less empathic and respectful responses, use of punitive sanctions and an overall reduction in adherence to reasonable standards of care (Bowers, 2002; Fraser & Gallop, 1993; Gallop et al., 1989; Watts & Morgan, 1994). Thus, prior to the last 20 years or so, the initial pic- ture regarding psychotherapy for PDs looked gloomy indeed: not only did research suggest that individuals with PDs were among the most treatment resistant patients, but also some clinicians took these fi ndings to heart and often responded to these patients with hopelessness or negativity.
Despite this worrisome beginning, recent evidence supports a much more optimistic outlook regarding the effi cacy and effectiveness of psychotherapy for individuals with personality disorders. In fact, a range of psychosocial treatments seem to be associ- ated with positive outcomes among those with PDs (Livesley, 2003). One review of 15 psychotherapy outcome studies revealed large effect sizes for self- and observer-rated outcomes in both naturalistic and randomized, controlled trials (RCTs) examining
psychotherapy for a variety of PDs (Perry, Banon, & Ianni, 1999). Further, a meta-analytic review of psychotherapy for individuals with Cluster C disorders revealed that these patients signifi cantly improve with cognitive behavioural therapy, psycho- dynamic therapy and social skills training, and these treatment gains are often main tained into follow-up periods of 3 months to 3 years (Simon, 2009). Finally, several RCTs using varied modes and styles of therapy have revealed promising improve- ments with psychotherapy for individuals with PDs (Arnevik et al., 2009; Bateman & Fonagy, 1999, 2008; Giesen-Bloo et al., 2006; Linehan et al., 2006; Svartberg, Stiles, & Seltzer, 2004). Given the evidence that hopelessness or pessimism regarding the ability to treat PDs can negatively impact the care that these patients receive (Bowers, 2002; Gallop et al., 1989; Fraser & Gallop, 1993; Watts & Morgan, 1994), it is crucial that mental health professionals familiarize themselves with this devel- oping literature.
Evaluating psychotherapy
Over the past several decades, increasing attention and effort has been directed toward understanding whether psychotherapy works. To address this ques- tion, researchers undertook a scientifi cally rigorous examination of the effects of psychotherapy for numerous mental disorders, with accumulating evi- dence pointing to positive effects for a variety of therapies (Smith & Glass, 1977). In 1995, Division 12 (Clinical Psychology) of the American Psycho- logical Association began a Task Force on Promo- tion and Dissemination of Psychological Procedures and a Task Force on Psychological Intervention Guidelines in order to further understand and promote the scientifi c support for psychological interventions. A few years later, Chambless and Hollon (1998) published comprehensive guidelines for established that a particular therapy is ‘ empiri- cally supported ’ . Briefl y, Chambless and Hollon (1998) note that, in order to be considered effi ca- cious, a therapy must have been shown to be benefi – cial in at least two carefully controlled studies. Specifi cally, the authors recommend putting the greatest weight on evidence derived from random- ized, controlled trials (RCTs), or trials in which par- ticipants are randomly assigned to receive the therapy in question or to a comparison condition (e.g. wait- ing list, treatment as usual, etc.), with the next great- est weight going to carefully controlled single-case or group experiments. In contrast, uncontrolled studies do not have comparison groups, and non- randomized studies involve non-random assignment to treatment conditions. Studies conducted as RCTs provide the strongest evidence that the observed
284 K. L. Dixon-Gordon et al.
effects are due to the therapy in question, and not to other confounding or common factors. Further, to reduce potential allegiance effects, evidence of a therapy ’ s benefi ts must be found by at least two inde- pendent researchers. If only one study is available to support the therapy or if the studies have all been conducted by the same research team, the therapy may be considered ‘ possibly effi cacious ’ . Chambless and Hollon (1998) also outline a number of metho- dological considerations that should be evaluated in order to determine that the data is of suffi cient qual- ity as to ensure confi dence in the conclusions, and provide guidelines for evaluating effi cacy when there are confl icting results.
In addition to evaluating a treatment ’ s effi cacy, Chambless and Hollon (1998) recommend the consideration of two other criteria: specifi city and effectiveness. To be considered specifi c, the therapy must be effi cacious and must demonstrate superior effects when compared with a control condition that incorporates the nonspecifi c processes of psycho- therapy, such as warmth, attention, expectation of change, and therapeutic rituals and rationales, among other factors (Wampold et al., 1997). A common way to evaluate specifi city is to compare the psycho- therapy under consideration to a ‘ treatment as usual ’ condition, often defi ned as treatment by existing programmes or resources in the community. A more stringent test of specifi city is to compare the therapy against another bona fi de or manualized treatment. For a therapy to be considered effective, a treatment must be shown to produce benefi ts in ‘ the real world ’ of clinical practice, where many of the stringent controls that characterize RCTs may not be present. In addition, Chambless and Hollon (1998) recommend that researchers and clinicians attend to the generalizability, feasibility and cost- effectiveness of therapy in clinical practice.
In this paper, we provide a review of the empirical literature evaluating psychotherapy for PDs. Using PsycINFO and Google Scholar, we searched for papers using the following key words on their own and in combination: personality disorder, Cluster A, Cluster B, Cluster C, psychotherapy, intervention, treatment, randomized controlled trial. Specifi cally, we have focused our review on evidence gleaned from randomized, controlled trials (RCTs) (see Table I). We excluded treatments developed for co-occurring diagnoses (e.g. treatments developed for an Axis I disorder and co-occurring PD). Where no RCTs were found, however, we provide a brief review of uncontrolled or case studies. Although not reviewed here, it is important to note that a wealth of naturalistic and case studies have evaluated therapy outcomes in PDs, also suggesting promise for various psychological approaches.
Evidence-based treatments for personality disorders
Unlike the Axis I psychotherapy literature, which often focuses on the evaluation of the effi cacy and effectiveness of particular therapeutic approaches within discrete diagnostic groups (e.g. CBT for depression, prolonged exposure for PTSD), much of the literature on psychotherapy for Axis II disorders examines the effi cacy of a therapeutic approach for individuals who meet criteria for a range of PDs. This approach makes intuitive sense, given that PDs have high rates of co-occurrence with other PDs (Conklin & Westen, 2005; Critchfi eld, Clarkin, Levy, & Kernberg, 2008; Hillbrand, Kozmon, & Nelson, 1996; Zanarini et al., 1998) and with Axis-I pathology (Skodol et al., 2002b; Zanarini et al., 1998). One notable exception to this trend is the case of borderline personality disorder (BPD), for which specialized treatment approaches have been devel- oped. Studies evaluating treatments to reduce crim- inal recidivism often include substantial proportions of individuals with antisocial personality disorder (ASPD), given the high prevalence of ASPD in offender populations (Hart & Hare, 1989); however, few studies have evaluated the treatment of ASPD specifi cally, and few studies of offender treatment explicitly separate fi ndings for those with ASPD versus those without (Duggan, Huband, Smailagic, Ferriter, & Adams, 2007). A few studies have also evaluated the effi cacy of psychotherapy for Cluster C disorders. Thus, we will fi rst review the evidence for the effi cacy of psychotherapy for per- sonality disorders in general; next, we present a review of the effi cacy of therapeutic approaches that have been developed to target-specifi c PDs.
Randomized controlled trials for mixed personality disorders
After a review of the empirical literature, we identifi ed six RCTs that have examined the effi cacy and specifi city of psychotherapy for mixed PDs (cf. Duggan et al., 2007). Across studies, the most common personality disorder diagnoses included borderline (24.4 – 44.4%), avoidant (5.4 – 40.3%), and obsessive – compulsive (16.2 – 37%). Less fre- quent diagnoses included schizotypal, schizoid, dependant, histrionic, antisocial, and narcissistic PDs, although there was considerable variability across studies. Among these studies, psychodynamic therapies were the most frequently examined thera- peutic approach.
Of these studies, two utilized waiting list control conditions (Huband, McMurran, Evans, & Duggan, 2007; Winston et al., 1994). In the fi rst
Psychotherapy for personality disorders 285 T
ab le
I .
R an
do m
iz ed
C on
tr ol
le d
T ri
al s
fo r
P er
so na
lit y
D is
or de
rs .
S tu
dy D
ia gn
os ti
c gr
ou p
S am
pl e
T re
at m
en t
C om
pa ri
so n
gr ou
p D
ur at
io n
P ri
m ar
y ou
tc om
es
M ix
ed P
D A
bb as
s, S
he ld
on ,
G yr
a, a
nd
K al
pi n
(2 00
8)
M ix
ed P
D
(4 4.
44 %
B P
D ,
37 %
O C
P D
, 33
.3 %
A P
D )
27 (
59 .3
% f
em al
e) ,
dr op
ou ts
n ot
kn
ow n
In te
ns iv
e sh
or t-
te rm
d yn
am ic
ps
yc ho
th er
ap y
(I S
T D
P )
M in
im al
c on
ta ct
co
nt ro
l M
� 2
7. 7
se ss
io ns
G lo
b al
s ev
er it
y (B
S I)
( in
f av
ou r
of I
S T
D P
); I
n te
rp er
so n
al
p ro
bl em
s (I
IP ),
( in
f av
ou r
of I
S T
D P
); G
lo b
al f
u n
ct io
n in
g (G
A F
) sy
m pt
om s,
s oc
ia l
(i n
fa vo
ur o
f IS
T D
P );
M ed
ic at
io ns
; H
ou rs
w or
ke d
( in
f av
ou r
of I
S T
D P
) A
rn ev
ik e
t al
. (2
00 9)
M ix
ed P
D 11
4 (7
4% f
em al
e) ,
11 %
d ro
po ut
D ay
h os
pi ta
l pr
og ra
m m
e,
14 %
d ro
po ut
O ut
pa ti
en t
in di
vi du
al
ps yc
ho th
er ap
y,
8% d
ro po
ut
18 w
ee ks
D ep
re ss
io n
(B D
I, B
H S
); G
lo ba
l se
ve ri
ty (
S C
L -9
0) ;
In te
rp er
so na
l pr
ob le
m s
(C IP
); G
lo ba
l fu
nc ti
on in
g (G
A F
)
H el
le rs
te in
e t
al .
(1 99
8) M
ix ed
P D
( 12
%
C lu
st er
C P
D ,
2% C
lu st
er A
, 3%
C lu
st er
B )
49 (
55 %
f em
al e)
, 34
.7 %
d ro
po ut
B ri
ef s
up po
rt iv
e ps
yc ho
th er
ap y,
29
% d
ro p
ou t
S ho
rt -t
er m
dy
na m
ic
ps yc
ho th
er ap
y,
40 %
d ro
p ou
t
40 s
es si
on s
G lo
ba l
se ve
ri ty
( S
C L
-9 0)
; P
at ie
nt c
om pl
ai nt
s (P
T C
); I
nt er
pe rs
on al
pr
ob le
m s
(I IP
)
H ub
an d
et a
l. (2
00 7)
M ix
ed P
D (
40 .3
%
A P
D ,
39 .2
%
B P
D ,
21 %
N
O S
)
17 6
(5 1.
1%
fe m
al e)
, 38
%
dr op
ou t
P sy
ch o-
ed uc
at io
n an
d pr
ob le
m
so lv
in g,
3 8%
dr
op ou
t
W L
, 0%
d ro
po ut
16 s
es si
on s
S oc
ia l
fu n
ct io
n in
g (S
F Q
); S
oc ia
l p
ro bl
em s
ol vi
n g
(S P
S I-