Therapy for Clients With Personality Disorders
lydash05_
Personalitydisorder.pdf
British Journal of Clinical Psychology (2015), 54, 129–146
© 2014 The British Psychological Society
www.wileyonlinelibrary.com
Personality disorder moderates outcome in short- and long-term group analytic psychotherapy: A randomized clinical trial
Steinar Lorentzen 1,2 *, Torleif Ruud
3,1 , Anette Fjeldstad
2 and
Per A. Høglend 1
1 Institute of Clinical Medicine, Clinic of Mental Health and Addiction, University of Oslo, Norway
2 Department of Research and Development, Clinic of Mental Health and Addiction, Oslo University Hospital, Oslo, Norway
3 Department for Research and Development, Division of Mental Health Services,
Akershus University Hospital, Lørenskog, Norway
Objectives. In a randomized clinical trial, short- and long-term psychodynamic
group psychotherapy (STG and LTG, respectively) schedules were equally effective
for the ‘typical’ patient during a 3-year study period. Although several studies have
reported good effects for patients with personality disorders (PD) in diverse forms
of psychotherapy, the significance of treatment duration is unclear. Therefore, we
tested the hypothesis that PD patients would improve more during and after LTG
than STG.
Design. A randomized, longitudinal, prospective study contrasting the outcomes during
and after short- and long-term dynamic group psychotherapies.
Methods. One hundred and sixty-seven outpatients with mood disorders, anxiety
disorders, or PD were randomized to STG or LTG (respectively, 20 or 80 weekly sessions
of 90 min each). Outcome measures are as follows: symptoms (SCL-90-R), interpersonal
problems (IIP-C), and psychosocial functioning (GAF split version: GAF-Symptom and
GAF-Function). PD pathology (number of PD criteria items) was selected a priori as a
putative moderator of treatment effects. Change during the 3-year study period was
assessed using linear mixed models. The study was registered at ClinicalTrials.gov as NCT
00021417.
Results. Our hypothesis was supported, as patients with PD improved significantly
more regarding all outcome variables in LTG than STG. For patients without PD, the rate
of change was similar across 3 years; however, the rate of change in symptoms and
interpersonal problems was higher in STG during the first 6 months.
Conclusions. The effectiveness of LTG is higher for patients with co-morbid PD.
Patients without PD do not appear to experience additional gain from LTG.
*Correspondence should be addressed to Steinar Lorentzen, Institute of Clinical Medicine, University of Oslo, Clinic of Mental Health and Addiction, P.O. Box 1039 Blindern, Oslo 0315, Norway (email: [email protected]).
DOI:10.1111/bjc.12065
129
Practitioner points Clinical implications
� LTG demonstrates better effectiveness than STG for patients with personality disorder co-morbidity (PD).
� Patients without PD do not appear to experience additional gain from attending LTG. � Correct initial allocation to treatment duration may prevent disruptive breaks in relationships and lead
to both human and economic cost savings.
Limitations
� Trials on mixed diagnostic samples may limit the ability to fully assess change for specific diagnostic groups.
� Therapists were unable to select patients and compose their own groups. Although this condition might increase the generalizability of the results, it may also have restricted the therapists and the
clinical situation inadvertently.
Meta-analyses and reviews indicate few differences in outcome between group therapies
with different theoretical rationale (Burlingame, Strauss, & Joyce, 2013; Lambert & Bergin,
1994). However, most group therapies studied are of short duration (<20 sessions; Burlingame et al., 2013), even though longer therapies (≥1 year) are common in clinical practice, at least in Europe (Lorentzen & Ruud, 2014; R€uger & Bell, 2004). Only a few studies demonstrate the effectiveness of long-term dynamic group psychotherapy with
outpatients (e.g., Bateman & Fonagy, 2009; Lorentzen, Bøgwald, & Høglend, 2002; Tschuschke, Anbeh, & Kiencke, 2007). Reviews of non-randomized studies that assess the
importance of treatment duration (Orlinsky, Grawe, & Parks, 1994; Orlinsky, Rønnestad,
& Willutzki, 2004) have failed to provide unequivocal answers. Because all observational
studies are open to several interpretations (Stiles & Shapiro, 1994), studies in which
patients are randomized to short- versus long-term group therapy are urgently needed.
Earlier studies have indicated that short-term therapies offer insufficient help to many
patients, including patients with personality disorders (PD) and patients with co-morbid
and chronic disorders (Gabbard, 2000; Lorentzen & Høglend, 2008; Perry, Banon, & Ianni, 1999; Piper, 1994; Reich & Green, 1991). However, some recent randomized clinical trials
have demonstrated more positive results of psychotherapy (including transference-
focused, mentalization -based, dialectical behaviour and schema-oriented therapies) with
patients that have various PD (Cluster C: Bartak et al., 2010; Svartberg, Stiles, & Seltzer,
2004; Vinnars, Barber, Noren, Gallop, & Weinryb, 2005; Mixed, including borderline PD:
Bateman & Fonagy, 2009; McMain, Guimond, Streiner, Cardish, & Links, 2012).
Interestingly, effective treatments have varied in duration from <20 weekly sessions to several years, rendering the studies inconclusive regarding the question of optimal treatment duration (Verheul & Herbrink, 2007). We are not familiar with any randomized
clinical studies that compare the effects of long-term and short-term group psychother-
apy.
A substantial proportion of outpatients meet criteria for several Axis-I and Axis-II
diagnoses and must cope with many long-term suboptimal functions, such as interper-
sonal problems, work problems, reduced physical health and higher overall mortality
(Holt-Lundstad, Smith, & Layton, 2010). Thus, we wanted to study outpatients within the
public mental health services; generally, these patients have been diagnosed with heterogeneous and co-morbid psychiatric disorders.
In this study, 6 months of group psychotherapy (20 weekly sessions) is compared with
2 years of group psychotherapy (80 weekly sessions) in a sample of outpatients, 48%
of which have one or more PDs (mostly Cluster C: avoidant, dependent, and obsessive-
130 Steinar Lorentzen et al.
compulsive). Outcomes over the 3-year study period were measured in the areas of
symptoms, interpersonal problems, and psychosocial functioning. Lorentzen, Ruud,
Fjeldstad, and Høglend (2013) reported recently that patients made significant gains
during a study period of 3 years in both STG and LTG, and they detected no significant longitudinal differences in change between patients in STG and LTG.
As indicated above, studies of psychotherapy that include populations of patients with
different PD have provided inconclusive answers regarding the significance of treatment
length. In this study, we test whether or not the treatment effect (difference in slopes
between LTG and STG) is significantly different between PD patients and non-PD patients.
Our hypotheses are that PD patients will do better in LTG than in STG, and that non-PD
patients will experience equivalent outcomes in the two treatments.
Method
Sites, patient referral, and inclusion
This study was conducted in three urban areas in Norway. One or two coordinators and
two to four group therapists were engaged at each site. The Data Inspectorate and The Regional Committee on Ethics in Health Research approved the project.
Mental health professionals and general practitioners were informed through mail of
the clinical trial; after evaluation, they referred patients they considered in need of
specialist services to the local coordinators, who assessed them for eligibility. Inclusion
criteria were one or more Axis-I or II diagnoses, interest in working with problems and
relationships in groups, and willingness to accept randomization. Exclusion criteria were
psychosis, main diagnosis of alcohol/drug addiction, organic brain disease, >6 months of previous group therapy, or psychotherapy ending within the last year. Written informed consent was obtained from all study participants.
Altogether, 175 patients were referred to the coordinators, and 167 eligible patients
were randomly allocated to the two treatment arms.
The local coordinator conducted the randomization for each site. The therapists did
not influence the randomization process (Schulz & Grimes, 2002). Stratified randomiza-
tion was performed for one short-term and one long-term group simultaneously, drawing
at least two women and two men for each group to secure mixed-sex groups.
The time interval from referral to assessment and randomization varied from 1 to 7 weeks. The time interval from randomization to the start of therapy varied from 2 to
12 weeks (mean interval 7 weeks). The coordinators were not involved in the treatment.
Groups, therapies and therapists
Eighteen psychotherapy groups were established that met for 90-min weekly sessions.
Nine of the groups were STG and nine were LTG. Our choice of therapy duration was a
compromise between enough time for the therapy to be effective, and how long we
expected most patients would be willing to participate. Each group consisted of seven to
eight patients and one therapist. Slightly more patients attended LTG, because three STGs
started with seven patients in order to prevent loss of patients. Besides, to ensure that a viable group would exist through the planned treatment period, protocol dictated that
patients that terminated in LTG during the first 6 months (n = 8 patients) were replaced by the next patient of the same sex. We wanted to avoid the high dropout rates and
collapse of therapeutic groups that some other group studies have experienced (e.g.,
Budman, Demby, Soldz, & Merry, 1996).
Personality disorder moderates therapy outcome 131
Both therapy formats were manualised group analytic therapies (Lorentzen, 2014)
built on the psychodynamic understanding of a developmental perspective on
personality, the existence of internal representations of interpersonal relationships,
psychological causation, and the influence of unconscious individual and group processes on behaviour. Facilitating or supportive techniques were used to promote interaction,
combined with traditional psychodynamic techniques of confrontation and interpreta-
tion (Foulkes & Anthony, 1984). The aim of group therapy was to become aware of
intrapsychic conflicts and dysfunctional interpersonal patterns, and to increase the
understanding of self, others and interpersonal relationships. Thus, the treatments studied
are not primarily targeted to specific symptom disorders. In STG, therapists were told to
be more active, to have a more circumscribed problem focus, to work more in the here
and now, and to be more attentive to the impending termination phase. Among the nine therapists (two men and seven women), there were two psychiatrists,
three psychologists, three psychiatric nurses, and one social worker. Their mean (SD)
age was 52.7 (3.7) years, time in practice 19.7 (4.4) years, and formal postgraduate
psychotherapy training 12.5 (3.7) years. Each therapist directed one short-term and one
long-term group and was trained in both formats. All therapists met regularly for
supervision.
Assessment
The pre-randomization evaluation (2–4 hr) included a full psychiatric record, the diagnostic interviews Mini International Neuropsychiatric Interview (MINI PLUS;
Sheehan et al., 2002), Structured Clinical Interview for DSM IV Axis-II Personality
Disorders (SCID-II; First, Gibbons, Spitzer, Williams, & Benjamin, 1997), and global
assessment of functioning (GAF; Endicott, Spitzer, Fleiss, & Cohen, 1976). The patients
also completed (as self-reports) the Symptom Check List-90-R (SCL-90-R; Derogatis, 1977)
and the inventory of interpersonal problems circumplex (IIP-C; Alden, Wiggins, & Pincus, 1990). After randomization, the patients met with their therapist individually for two
sessions, during which they received more information about the group and negotiated a
treatment contract.
The self-report outcome measures (SCL-90-R, IIP-C) were completed for all patients
every 6 months in both treatment formats for as long as the LTG was running and for
12 months after LTG termination. Self-reports were also sent to non-starters premature
enders. Selected sessions from both therapies (early, mid-, and end phases) were audio-
recorded. The same coordinators interviewed the patients 3 years after the beginning of
therapy; at this point, the patients were diagnosed on Axes I and II and rated on GAF. We
consider this assessment to be ‘close to blind’, because the coordinators could not
possibly remember the original allocation of the patients, who were instructed to
conceal this.
Measures
Outcome measures
Because most patients had high co-morbidity and chronicity, we selected several
measures intended to capture the diversity of change in symptomatic distress, interpersonal problems, severity of psychopathology, and psychosocial functioning.
The main targets for interventions during group psychotherapy are interpersonal
132 Steinar Lorentzen et al.
problems and psychosocial functioning; these are also the primary outcome variables in
this study.
Symptom checklist-90-revised. The SCL-90-R (Derogatis, 1977) consists of 90 items
rated on a 5-point Likert scale. The general symptom load is reflected in the Global
Severity Index (GSI), which is the mean of all ratings. Cronbach’s alpha was .96 in this
study.
Inventory of interpersonal problems-circumplex. The IIP-C (Alden et al., 1990) is well
validated, and is one of the most widely used self-reports in psychotherapy research. Sixty- four items are rated on a 5-point Likert scale. Cronbach’s alpha was .94 in this study. The
IIP has demonstrated high test–retest stability across 10 weeks, but is still sensitive to change (Horowitz, Rosenberg, Baer, Ureno, & Villasenor, 1988).
Global assessment of functioning (Axis V, DSM-IV; Endicott et al., 1976). Symptom
severity and level of functioning were assessed using the split version of the GAF scale
(Goldman, Skodol, & Lave, 1992), with separate scores for symptom severity (GAF-S) and problems in functioning (GAF-F). The intraclass correlation coefficient (ICC; two-way
random) among five evaluators rating 20 cases was 0.78 for the GAF-S and 0.80 for the GAF-
F (Shrout & Fleiss, 1979).
Target complaint. The target complaint (Battle, Imber, Hoehn Saric, Nash, & Frank,
1966) is a self-report measure of the problems for which a person seeks therapy. This
measure was used as a measure of expectancy in this study. Each patient described the primary problem and rated its severity on a scale ranging from 1 = no problem to 12 = could not be worse, and Expectation of Change from 1 = change to the worse to 12 = go away completely. The duration of target complaint was also recorded.
Moderator
Presence of PD pathology
Axis-II diagnoses (PD criteria items; DSM-IV) were established by the local coordinators
using SCID-II at baseline (First et al., 1997). According to the DSM-IV, the PD comprises 10
specific disorders and one non-specific disorder (American Psychiatric Association [APA], 1994).A consistent finding is that patients with one specific PD often exhibit a high degree
of co-morbidity with other PDs (Critchfield & Benjamin, 2006). The number of fulfilled PD
criteria items has a linear relationship with quality of life and interpersonal dysfunction,
independent of specific PDs (Cramer, Torgersen, & Kringlen, 2006). In agreement with
the view that cumulative scores of criteria items for PD represent the data better than
categorical scores (Gunderson, Links, & Reich, 1991; Hersoug, Monsen, Havik, &
Høglend, 2002; Levenson, Wallace, Fournier, Rucci, & Frank, 2012), we decided to focus
on overall personality pathology rather than specific PDs. The overall personality pathology is the sum of positive criteria items on the SCID-II. The scores ranged from 0 to
31 in this study, with a mean of 8 (6.8) criteria items.
Personality disorder moderates therapy outcome 133
Statistical analysis
With a sample of 120 patients, standard power calculation (end-point analysis) indicated
that a moderate effect size (0.5) could be detected for an alpha threshold of 0.05 with a
power of 0.80 (Altman, 1991). We aimed to include 170 patients, to ensure an adequate sample size of completers in the event of substantial drop-out rates. We used linear mixed
models (LMMs) to analyze longitudinal data (SPSS, 2008). Use of a LMM allows non-
independence of repeated and nested data. It can handle incompleteness due to missing
data. To allow for non-independence of data within patients nested within groups nested
within therapists, ‘patient’, ‘group’ and ‘therapist’ were each treated as random effects:
randomly distributed intercepts and slopes were fitted for each patient, for each group,
and for each of the nine therapists. Treatment (group duration) was coded LTG = 1, STG = 0. Time was coded as a time interval axis, with months as the unit. When the SCL- 90-R or the IIP was used as the outcome variable, a two-piece linear model with the knot at
time = 6 (the end of STG) fitted the data discernibly better than did a linear time slope (change in – 2 log likelihood, p < .005).
To test the treatment effects over time (difference in slopes between STG and LTG), as
well as the putative moderator effect of PD pathology, the following composite model
equation was used: Yij = B0 + B1TIMEij + B2 (TIMEij 9 TREATMENTi) + B3PDi + B4(PDi 9 TIMEij) + B5(PDi 9 TIMEij 9 TREATMENTi) + [foi + f1iTIMEij + eij]. Yij is the dependent variable score for patienti at timepointj; B0–B5 are the fixed effects; and foi, f1iTIMEij, and eij are random intercept, random time and the error term, respectively. Random effects for groups nested within therapists were also tested. By design, treatment
group means were equal at baseline. The statistical model forces both treatments to have a
common intercept. This model is more powerful and is routinely recommended for the
analysis of randomized clinical trials (Fitzmaurice, Laird, & Ware, 2004). After entering
time at level one, all other predictors were entered at level two.
The relevant parameter is the treatment effect, B2 (time 9 group length). That is, the
difference in slopes between LTG and STG. The moderator effect, B5 (time 9 group length 9 PD pathology), signifies whether or not the treatment effect differs for patients
with high versus low levels of PD pathology. Moderators should not be determined by p-
values, but rather by effect sizes (Kraemer, Wilson, Fairburn, & Agras, 2002). When PD
pathology is centred at the mean value among patients with PD (PD criteria items = 12), the treatment effect can be interpreted directly as the effect for the typical (average) PD
patient. When PD pathology is centred at the mean value among patients without PD (PD
criteria items = 3), the treatment effect can be interpreted directly as the effect for the typical non-PD patient.
No longitudinal analyses were conducted on subgroups of patients. Effect sizes
(converted to Cohen’s d), derived from the F-test for mixed effects model, were calculated
as d ¼ 2 ffiffiffiffi F df
q , where F is the F-test statistic for the effect of interest in the repeated model,
as well as other multilevel designs (Verbeke & Molenberg, 2000). Dropout status, defined
as attending less than two-thirds of the scheduled sessions (i.e., fewer than 53 LTG
sessions or 13 STG sessions), was based on clinical consensus and determined before the
statistical analyses were run. Non-starters were randomized patients that did not appear
for group therapy. A pattern mixture approach was used to explore whether or not missing data influenced the results (Gibbons & Hedeker, 1997). That is, we included
starter status and dropout status in the statistical models. The following terms were
entered: status, status 9 time, and status 9 time 9 treatment.
We also calculated the clinically significant change for patients who started in
therapy. Reliable change (RC) is defined as a change that exceeds what could be
134 Steinar Lorentzen et al.
expected from error variance alone (i.e., there is a 95% probability of a true change). It
is estimated for each patient (who is rated to be in the dysfunctional area), on every
outcome measure, based on the reliability of the measurements (Jacobson & Truax,
1991). To obtain clinically significant change (recovery), the patients must also cross a cut-off score between a clinical and a non-clinical sample (Jacobson & Revenstorf,
1988); we used norms from a Norwegian outpatient population to calculate this
threshold (Monsen, Kallerud, Eilertsen, & von der Lippe, 1999). Patients that
improve reliably, without moving into a more healthy population, are classified as
‘improved’, while patients who change reliably in a negative direction are classified as
‘deteriorated’. When the GSI and IIP were used as outcome variables, a change of 0.29
and 0.35 scale points represented RC, and we calculated cut-off points between
dysfunctional and functional populations to be 0.65 and 1.1, respectively (Jacobson & Revenstorf, 1988). For GAF-S and GAF-F, RC was 7.0 scale points, with a cut-off point of
70 on the GAF scale.
We also calculated the number of patients that needed treatment, in order to
implement a clinically significantly change in one LTG patient with PD, compared to an
STG patient: number needed to treat (NNT) = 1/ARR (absolute risk reduction) = 1/CER (control event rate) � EER (experimental event rate). CER and EER are the percentages of patients that have achieved a clinically significant change in STG and LTG, respectively
(see above).
Results
Patient flow
A flow chart describing the patient dispensation is depicted in Figure 1.
Although a total of 175 patients were evaluated, eight patients did not meet the study
inclusion criteria. One hundred sixty-seven patients were randomized (90 to LTG and 77
to STG). Nineteen patients did not appear for the group (12 in LTG and 7 in STG; v2 = 0.72, df = 1, p = .47). We have information from 13 of these 19 patients (68%) regarding why they did not start in therapy: four chose alternative treatments while
waiting for the group to start, six were sceptical of their therapist or the group format, and
three had external reasons (change in employment). There was no difference in positive
expectations of treatment between these patients and starters, t(164) = 0.57, p = .57, 95% CI [�0.89, 1.60].
Rates of premature termination differed: 26 patients (33.3%) in LTG and 6 patients
(8.6%) in STG, v2 = 11.9, df = 1, p < .001. The number of premature terminations during the first 6 months was equal in STG and LTG, as we demonstrated previously in
a Kaplan–Meier diagram (Lorentzen et al., 2013). After 6 months, the attrition per time unit remained the same in LTG as during the first 6 months. We have information
about why 26 out of 32 patients (81%) ended prematurely: ten thought the group was
not helpful, eight were dissatisfied with the group or the therapist for different reasons,
four obtained the help they needed, and four had external reasons (usually change in
employment).
The mean number of sessions in treatment for starters was 18.9 (3.8) in STG and 57.7 (26.9) in LTG, which comprises respective attendance rates of 94.5% (19.1) and 72.1%
33.9; t(122) = 5.0, p < .0005 (independent t-test, two-tailed). Seventy-nine of 90 patients (88%) completed the 3-year follow-up interview after LTG versus 71 of 77 patients (92%)
after STG. For the PD patients (n = 75), 91% and 93% participated in the follow-up evaluations after LTG and STG, respectively.
Personality disorder moderates therapy outcome 135
Patient characteristics at baseline
Table 1 summarizes the patients’ baseline characteristics, for the whole sample
(n = 167). There were significantly more patients with generalized anxiety disorder in LTG (v2 = 3.84, df = 1, p = .05). All other differences were statistically non-significant.
The mean number of co-morbid Axis-I diagnoses per patient was 3.3. Patients reported
that the problems they wanted therapy for had a mean duration of 15 years. Sixty-eight percent reported previous psychiatric treatment.
Correlations between outcome variables are presented in Table 2. With the exception
of psychosocial functioning, most correlations were small to moderate, indicating that
they measured different aspects of change.
Referrals to coordinator and assessed for eligibility n = 175
Excluded, did not meet inclusion criteria n = 8
Included in random assignment n = 167
Assigned LTG n = 90
Assigned STG n = 77
Started treatment n = 78
Started treatment n = 70
Withdrew N = 12 (13.3%)
Withdrew N = 7 (9.1%)
Received interven on n = 52 (57.8%)
Discon nued N = 26 (33.3%)
Received interven on n = 64 (83.1%)
Discon nued n = 6 (8.6%)
Completed 3-year follow-up interview n = 71 (92 %; includes 9 withdrawals and premature termina ons)
Completed 3-year follow-upinterview n = 79 (88 %; includes 30 withdrawals and premature termina ons)
Figure 1. Patient flow in a randomized clinical trial of short- and long-term group analytic
psychotherapy.
136 Steinar Lorentzen et al.
Therapist and group effects
With the GSI and IIP-C as outcome variables, random variances for groups and therapists
were small; all ICCs were <.02. When the GAF-S was used as an outcome variable, the random intercept variance for therapists indicated some non-independence in the data (ICC = .02). The therapist intercept was therefore included as a random effect in the statistical model. This improved goodness-of-fit significantly (change in �2LL, p < .01). When the GAF-F was employed as an outcome variable, the random intercept variance for
therapists also indicated some non-independence in the data (ICC = .02), and therapist intercept was included in the statistical model as a random effect. It should be noted that
Table 1. Pretreatment characteristics of randomized patients in short- and long-term group analytic
psychotherapy (n = 167)
Short-term groups Long-term groups Sample
n = 77 n = 90 n = 167 Mean (SD) Mean (SD) Mean (SD)
Age 38.6 (9.4) 38.2 (9.4) 38.4 (9.4)
Education 13.8 (3.3) 13.7 (3.1) 13.7 (3.2)
Expectations a
7.9 (2.4) 7.2 (2.7) 7.7 (2.6)
Number axis-I diagnoses 3.2 (1.9) 3.4 (2.0) 3.3 (2.0)
Number axis-II diagnoses 0.5 (0.9) 0.7 (0.9) 0.6 (0.9)
SCID-II (DSM-IV) Criteria 7.5 (7.3) 8.5 (6.4) 8.0 (6.8)
n (%) n (%) n (%)
Female 50 (65) 55 (61) 105 (63)
Single 34 (44) 41 (46) 75 (45)
Formerly hospitalized 7 (9) 13 (14) 20 (12)
Axis-I diagnosis
Major dep, single 5 (7) 10 (12) 15 (10)
Major dep, recurrent 53 (69) 61 (72) 114 (68)
Dysthymia 7 (9) 3 (20) 10 (6)
Panic disorder 23 (30) 34 (40) 57 (34)
Agoraphobia 8 (10) 4 (4) 12 (7)
Obsess-Comp disorder 8 (10) 20 (21) 28 (17)
Social phobia 24 (31) 31 (38) 55 (33)
GAD 11 (14) 28 (30) 39 (23)
PTSD 6 (8) 1 (1) 7 (4)
Somatoform disorder 12 (16) 23 (26) 35 (21)
Other 29 (38) 28 (31) 57 (34)
No axis-I diagnosis 1 (1) 3 (3) 4 (2)
Axis-II diagnosis
Avoidant 18 (23) 29 (32) 47 (28)
Dependent 3 (4) 5 (6) 8 (5)
Obsessive compulsive 3 (4) 11 (12) 14 (8)
Paranoid 7 (9) 4 (4) 11 (7)
Borderline 5 (7) 4 (4) 9 (5)
PD NOS 4 (5) 9 (10) 13 (8)
Antisocial 0 2 (2) 2 (1)
Note. GAD = generalized anxiety disorder; PTSD = post traumatic stress disorder; SCID-II (DSM- IV) = Structural Clinical Interview for DSM-IV Axis-II Personality Disorders. a After randomization. v2 = 3.84, p = .05 (n = 167).
Personality disorder moderates therapy outcome 137
ICCs for groups might be underestimated because each therapist provided only two
groups (one in each condition), which also confounds the group and therapist effects in
this study.
Treatment fidelity
Thirty-nine audio recordings from the first 6 months of therapy (sessions 3, 10, and 17 of
both formats) were drawn from the pool of 54 recordings. Two evaluators blindly and independently rated sessions on therapist activity level, degree of focus, group work in the
here and now, and therapist competence, using Likert scales from 0 (not at all) to 4 (very
much). ICC on the therapy process scales ranged from 0.70 to 0.94. There was significantly
more workona circumscribedproblem focus– t(15) = 2.3, p = .04 – and work in the here- and-now in STG – t(35) = 2.1, p = .04, as suggested in the treatment manuals. The levels of therapist activity were equal in the two groups. Therapist mean competence was similar in
both formats, ranging from moderate to high: M = 2.7 (0.5), range 1.7–3.4, in STG; and 2.3 (0.5), range 1.9–3.1, in LTG. We reported previously that there were no significant differences (sessions 3, 10, and 17) in therapeutic alliance or group cohesion between the
two group formats (Bakali, Wilberg, Hagtvet, & Lorentzen, 2010).
Moderator analyses
One case was deleted from all analyses of longitudinal data. This patient did not meet study
inclusion criteria and should not have been included; however, the main reason for
exclusion was that the case was an extreme outlier, >3 SDs away from the mean of the distributions. Including this case significantly worsened goodness-of-fit measures (change
in �2 log likelihood, p < .001). Therefore, all longitudinal analyses were performed on 166 patients.
The interaction term PD 9 time 9 treatment tests whether or not PD pathology
moderates the treatment effect (that is, whether the treatment effect is different for
patients with high vs. low PD pathology). When we used the GSI as an outcome measure,
we were not able to demonstrate a significant moderator effect of PD pathology during the
first 6 months. When we used the IIP, the treatment effect tended to be stronger for patients without PD pathology during the first 6 months. Although the difference was not
statistically significant, the effect size of this moderator term was 0.3. When the GSI or IIP
were used as outcome variables, PD pathology did not moderate the treatment effects
during the last 2.5 years. Both patients with and without PD pathology improved more in
LTG than in STG during this period. Using the GAF-S, we were able to detect a moderator
Table 2. Intercorrelations between outcome variables at baseline
1 2 3 4
Symptoms, GSI .55** �.30** �.38** Interpers. prob. IIP-C .55** �.11 �.18* GAF-symptoms �.30** �.11 .75** GAF-function �.38** �.18* .75**
Note. GAF = global assessment of functioning; IIP-C = inventory of interpersonal problems circumplex; GSI = Global Severity Index. *Significant at .05, and **significant at .01 level (2-tailed).
138 Steinar Lorentzen et al.
effect of PD pathology, b = .014 – F(1, 148) = 6.34, p = .06; the effect size was 0.4. With the GAF-F as an outcome variable, the trend moderator effect was b = .013 – F(1, 148) = 3.13, p = .08; the effect size was 0.3. In both cases change was measured across 3 years, and patients with greater PD pathology changed more.
When we included starter status (1, 0) and/or premature termination status in the
statistical models (Gibbons & Hedeker, 1997), these terms were non-significant,
goodness-of-fit (�2LL) was not improved, and the pattern of results did not change. These results were to be expected because the proportions of missing data were
low.
Typical patients with or without PD For the typical (average) patient with PD (PD criteria items = 12), we did not detect any significant differences in effectiveness between LTG and STG during the first 6 months
of the study period when we used the GSI and IIP as outcome variables. The parameters
were b = .01, F(1, 217) = 0.22, p = .64 and b = .11, F(1, 211) = 0.79, p = .38, for GSI and IIP, respectively. However, during the last 2.5 years, LTG was superior to STG, as
measured with both the GSI and the IIP. For the GSI, b = �.01, F(1, 229) = 10.24, p = .002; the effect size was 0.4. For the IIP, b = �.004, F(1, 218) = 3.74, p = .05; the effect size was 0.3. When the GAF-S was employed as an outcome variable, LTG worked significantly better than STG. The treatment effect was b = .15, F(1, 144) = 6.17, p = .01; the effect size was 0.4. When the GAF-F was employed as an outcome variable, there was a trend that LTG was superior to STG: b = .11, F(1, 144) = 3.5, p = .07, with an effect size of 0.3.
For the typical patient without PD (PD criteria items = 3), there were no differences in effectiveness between the two treatments, when the GAF-S and GAF-F
were employed as outcome variables. When the GSI was used as an outcome measure,
there was no significant difference between STG and LTG during the first 6 months – b = .02, F(1, 204) = 2.43, p = .12, although STG tended to exhibit more improvement than LTG. The effect size was 0.2. From 6 to 36 months, there was a therapy effect in
favour of LTG: b = �.01, F(1, 227) = 14.75, p < .005, with an effect size of 0.5. When the IIP was used, STG worked significantly better than LTG during the first 6 months,
with a corresponding increased change in LTG from 6 to 36 months. The respective
parameters are b = .04, F(1, 197) = 9.91, p = .002, effect size 0.4; and b = �.01, F(1, 210) = 7.8, p = .006, effect size 0.4.
Clinically significant and reliable change
Among PD patients, clinically significant change (recovery) as measured by the GSI
was achieved by 41.4% of patients in LTG versus 26.3% of patients in STG (v2 = 0.58, df = 1, p = .45). For the IIP, the corresponding numbers were 41.2% versus 21.1% (v2 = 1.4, df = 1, p = .24). When the GAF-S was employed as an outcome measure, recovery was achieved by 44.7% in LTG versus 18.2% in STG (v2 = 3.51, df = 1, p = .06). When the GAF-F was employed, the corresponding figures were 54.3% versus 42.9% (v2 = 0.44, df = 1, p = .51). All calculations are with continuity correction.
The NNT in order to have one more PD patient recovered in LTG than STG is
(measured for each outcome variable): 7 for the GSI, 5 for the IIP, 4 for the GAF-S and 9 for
the GAF-F.
Personality disorder moderates therapy outcome 139
Differential changes in PD criteria for Clusters A, B and C
Preliminary secondary analyses indicate that endorsed criteria for Clusters C and A are
more important moderators of change than Cluster B criteria. This finding will be
explored further in future studies.
Deterioration
We did not detect a significant difference between STG and LTG regarding the
proportion of patients that deteriorated. Across all outcome variables, 5.7% deterio-
54
56
58
60
62
64
66
68
70
72
74
–2 36
G A F- F
Months Short-term PD + (N = 28) Long-term PD + (N = 46) Long-term PD – (N = 43)
Short-term PD – (N = 49)
54
56
58
60
62
64
66
68
70
72
74
–2 36
G A F- S
Months
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
0 6 12 18 24 36
G SI
Months
Short-term PD + (N = 25) Long-term PD + (N = 44) Long-term PD – (N = 38)
Short-term PD – (N = 46)
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
0 6 12 18 24 36
IIP
Months
(A)
(C) (D)
(B)
Figure 2. Change in the four main outcome variables during 3 years. (A) Global Severity Index (GSI)
score for patients with and without personality disorders (PD) in long-term and short-term therapy
(N = 153). (B) Inventory of interpersonal problems (IIP) score for patients with and without PD in long- term and short-term therapy (N = 153). (C) global assessment of functioning (GAF) symptom score for patients with and without PD in long-term and short-term therapy (N = 166). (D) GAF function score for patients with and without PD in long-term and short-term therapy (N = 166).
140 Steinar Lorentzen et al.
rated in STG versus 5.9% in LTG. The corresponding proportion of deteriorators among
PD patients was 4.4% versus 7.0% among patients without PD. Burden and other
adverse effects for patients were small. There were no suicide attempts or completed
suicides. Figure 2, presents the descriptive change regarding all outcome variables during the
entire study period. LTG and STG exhibited similar effectiveness for patients with PD
during the first 6 months (Figure 2A,B). After 6 months, PD patients in STG improved
very little, while PD patients in LTG continued to improve. For patients without PD, STG
appeared to work faster initially. When the IIP was employed as the outcome variable,
LTG participants did not quite ‘catch up’ after 36 months. PD patients improved more in
LTG than in STG (Figure 2C,D). The improvement rate was almost identical for patients
without PD. The patients were interviewed about additional treatment (e.g., hospitalizations,
psychoactive drugs, consultations, etc.) and sick leave during the 3 years. Surprisingly,
there were not any statistically significant differences between STG and LTG regarding the
use of health services by patients with or without PD. Therefore, the use of additional
treatment was unlikely to be a confounding factor in this study.
The patients recorded positive and negative life events annually (before therapy start,
and at 1, 2, and 3 years). There were no significant differences between STG and LTG for
patients with or without PD.
Discussion
As far as we know, this is the first randomized clinical trial to compare the effectiveness of
psychodynamic short- and long-term group psychotherapy. We reported earlier that we
were unable to detect any difference in effectiveness, on average, between the two
treatment arms throughout the 3-year study period (Lorentzen et al., 2013).
However, LTG was more effective than STG for the typical PD patient (PD criteria items = 12). When the SCL-90-R and IIP were used as outcome variables, improvement in the two groups was similar during the first 6 months, but LTG was more effective during
the last 2.5 years of the study period. When the GAF-S and GAF-F were employed as
outcome variables, LTG was more effective than STG.
In this study we have observed moderator effects of PD pathology, as demonstrated by
the significantly higher change of GAF-S and GAF-F scores among patients with greater PD
pathology that participated in LTG versus STG.
For the typical non-PD patient (PD criteria items = 3), we did not detect any significant difference in effectiveness over the 3-year study period. When the SCL-90-R and IIP were
used as outcome variables, patients without PD tended to improve more in STG than in
LTG during the first 6 months of the study period. This is probably due to the time limit,
the more circumscribed problem focus, and more work in the here and now, which
appears to quicken the initial change process. However, these patients changed
significantly more in LTG than in STG during the last 2.5 years. As it may seem that the STG
and LTG patients end up equally well after 36 months, the results may be difficult to
interpret. However, because STG tends to work faster than LTG, it could be argued that STG tends to be more favourable than LTG for non-PD patients. A follow-up study at a later
time-point might indicate the degree of further change in STG and LTG.
There was a greater number of premature terminations among participants in LTG
than STG, but no increase in attrition rate throughout LTG. This finding means that
the greater number of dropouts in LTG is partly due to the fact that LTG is four times
Personality disorder moderates therapy outcome 141
as long as STG. However, in retrospect there also seemed to be greater dissatisfaction
among therapists regarding group composition in LTG, and a few therapists also
failed to integrate some members into their groups (Horneland, Sande, Høbye,
Knudsen, & Lorentzen, 2012). Groups formed by randomization might have had a suboptimal composition compared with the groups that the therapists would have
put together.
Surprisingly, the STG patients did not seek more additional treatment than the patients
in LTG during the 3-year study period. As part of the clinical routine (and recommendation
in the manual) at the end of therapy the patients were asked not to seek additional
treatment during the following year unless treatment was urgently needed, because
delayed positive effects after treatment might be experienced. This advice may have been
easier to follow because patients in STG (especially those without PD) were also improved after therapy.
Although the proportion of clinically significant change (recovery) among PD patients
was higher among LTG participants than among STG participants, we were unable to
demonstrate significant differences (except for the GAF-S), which may be caused in part
by a low n (Type II error). Still, we think that the larger treatment effect in the LTG group
justifies a longer treatment for most PD patients. In the largest meta-analysis of active
psychotherapy versus placebo/waitlist control (Lipsey & Wilson, 1993), NNT = 4. It should be noted that our study compares two active treatments, and that STG patients also undergo important change. In spite of this, our NNT among PD patients is about as strong
as treatment versus placebo control for two of the outcome variables. PD patients may be
difficult to treat (Bender et al., 2001; Perry et al., 1999), and they experience a high
burden of disease, comparable to severe somatic illnesses (Soeteman, Verheul, &
Busschbach, 2008). Recovery/improvement in PD patients may therefore represent both
economic and human cost savings.
Strengths and limitations
This study has several strengths. First, the treatment took place in ordinary treatment
settings (outpatient departments and private practices) with regular patients and
therapists, which may allow us to generalize our findings to these clinical situations. Other
strengths are the high inclusion rate from referral and the regular sending of self-reports to
all randomized patients. This, and the fact that approximately 90% of the patients
participated in a 3-year follow-up, has resulted in a nearly complete data set, allowing for
reliable results from the statistical analyses. Despite these strengths, several limitations should also be noted. Although
randomization is the ‘gold standard’ from a methodological point of view, in some
respects it may represent a drawback from a clinical point of view. Therapists were not
allowed to select their own patients and ‘compose’ their groups in the way they usually
would, which may restrict the generalizability of our findings to some extent
(Horneland et al., 2012). On the other hand, therapists were not allowed to exclude
patients included in the study, which may in fact increase generalizability, and might
reduce the magnitude of treatment effects (and thereby reduce the likelihood of overestimating them). The patient sample had a wide variety of diagnoses, which may
make it difficult to evaluate the effects of group therapy duration on a specific
psychiatric disorder, such as depression. On the other hand, the wide variety of
diagnoses may actually increase generalizability to patients seeking outpatient group
psychotherapy.
142 Steinar Lorentzen et al.
Clinical implications
Patients entering group therapy should be carefully evaluated for the presence of PD
before a specific duration of therapy is prescribed. A routine of offering all patients
LTG, which is not uncommon in slow-open, mixed groups, appears to be unnecessary and may be a waste of resources. A routine of offering all patients STG, and continuing
in LTG if STG provides insufficient improvement, may be an alternative. However,
changing groups and therapists can also entail problems, probably more so for patients
with PD. Correct initial allocation of patients may prevent disruptive breaks in
therapeutic alliances and bonding to fellow group members among PD patients, and
may also protect patients without PD from a long period of less productive therapy.
Careful initial evaluation and patient allocation may lead to both economic and human
cost savings, as well as more realistic expectations of improvement within specific time frames.
Conclusion
This study indicates that LTG is more favourable for patients with co-morbid PD. For patients without PD, LTG does not provide any additional gain. Replication of these results
is needed. A new trial that examines STG or LTG for PD patients only would provide a
more definitive test of the hypothesis generated in this study.
Acknowledgements
We want to express gratitude to the following Institutions and persons for important
contributions: Outpatient clinics at Community Mental Health Centers in�Alesund, Alna (Oslo),
in Sandnes and Ryfylke (Stavanger), Laila Hjulstad, Knut Skjøstad, Vibeke Lohne, Allan Larsen,
Kirsten Høbye, Helge Knudsen, Marthe Horneland, Dagny Sande Børnes, Ole Inge Gjøen,
Wibeke Kløvning, Synnøve A. Kristiansen, Martin Mydske Nilssen, Ørjan Berg.
References
Alden, L. E., Wiggins, J. S., & Pincus, A. L. (1990). Construction of circumplex scales from the
inventory of interpersonal problems. Journal of Personality Assessment, 55, 521–536. doi:10. 1080/00223891.1990.9674088
Altman, D. G. (1991). Practical statistics for medical research. London, UK: Chapman & Hall.
American Psychiatric Association (1994). Dianostic and Statistical Manual of Mental Disorders
(DSM-IV). Washington, DC: Author.
Bakali, J. V., Wilberg, T., Hagtvet, K. A., & Lorentzen, S. (2010). Sources accounting for alliance and
cohesion at three stages in group psychotherapy: Variance component analyses. Group
Dynamics: Theory, Research, and Practice, 14, 368–383. doi:10.1037/a0019170 Bartak, A., Spreeuwenberg, M. D., Andrea, H., Holleman, L., Rijnierse, P., Rossum, B. V., . . .
Emmelkamp, P. M. (2010). Effectiveness of different modalities of psychotherapeutic treatment
for patients with cluster c personality disorders: Results of a large prospective multicentre study.
Psychotherapy and Psychosomatics, 79, 20–30. doi:10.1159/000254902 Bateman, A., & Fonagy, P. (2009). Randomized controlled trial of outpatient mentalization-based
treatment versus structured clinical management for borderline personality disorder. The
American Journal of Psychiatry, 166, 1355–1364. doi:10.1176/appi.ajp.2009.09040539 Battle, C. C., Imber, S. D., Hoehn Saric, R., Nash, E. R., & Frank, J. D. (1966). Target complaints as
criteria of improvement. American Journal of Psychotherapy, 20, 184–192. Retrieved from http://www.ajp.org/
Personality disorder moderates therapy outcome 143
Bender, D. S., Dolan, R. T., Skodol, A. E., Sanislow, C. A., Dyck, I. R., McGlashan, T. H., . . . Gunderson, J. G. (2001). Treatment utilization by patients with personality disorders. The American Journal
of Psychiatry, 158, 295–302. doi:10.1176.appi.ajp.158.2.295 Budman, S. H., Demby, A., Soldz, S., & Merry, J. (1996). Time-limited group psychotherapy for
patients with personality disorders: Outcomes and drop-outs. International Journal of Group
Psychotherapy, 46, 357–377. Retrieved from http://www.guilford.com/cgi-bin/cartscript.cgi? page=pr/jngr.htm
Burlingame, G. M., Strauss, B., & Joyce, A. S. (2013). Clinical mechanisms and effectiveness of small
group treatment. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook of psychotherapy and
behavior change (5th ed., pp. 640–689). New York, NY: Wiley. Cramer, V., Torgersen, S., & Kringlen, E. (2006). Personality disorders and quality of life.
A population study. Comprehensive Psychiatry, 47, 178–184. doi:10.1016/j.comppsych.2005. 06.002
Critchfield, K., & Benjamin, L. S. (2006). Principles for psychosocial treatment of personality
disorder: Summary of the Division 12 Task Force review. Journal of Clinical Psychology, 62,
661–674. doi:10.1002/jclp.20255 Derogatis, L. R. (1977). The SCL-90-R: Administration, scoring and procedures. Manual I.
Baltimore, MD: Clinical Psychometric Research Unit, John Hopkins University School of
Medicine.
Endicott, J., Spitzer, R. L., Fleiss, J. L., & Cohen, J. (1976). The Global Assessment Scale: A procedure
for measuring overall severity of psychiatric disturbance. Archives of General Psychiatry, 33,
766–771. doi:10.1001/archpsyc.1976.01770060086012 First,M. B., Gibbons, M., Spitzer, R. L., Williams, J. B.W., &Benjamin, L. S. (1997). Structural Clinical
Interview for DSM-IV Axis II Personality Disorders (SCID-II). Washington, DC: American
Psychiatric Press.
Fitzmaurice, G., Laird, N. M., & Ware, J. H. (2004). Applied longitudinal analysis. New York, NY:
Hoboken.
Foulkes, S. H., & Anthony, E. J. (1984). Group psychotherapy. The psychoanalytical approach.
London, UK: Maresfield Reprints. First published London, UK: Penguin Books, 1957.
Gabbard, G. O. (2000). Psychotherapy of personality disorders. The Journal of Psychotherapy
Practice and Research, 9, 1–6. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/ PMC3330582/
Gibbons, R. D., & Hedeker, D. (1997). Random effects probit and logistic regression models for
three-level data. Biometrics, 53, 1527–1537. doi:10.2307/2533520 Goldman, H. H., Skodol, A. E., & Lave, T. R. (1992). Revising axis V for DSM-IV: A review of measures
of social functioning. The American Journal of Psychiatry, 149, 1148–1156. Retrieved from http://ajp.psychiatryonline.org/journal.aspx?journalid=13
Gunderson, J. G., Links, P. S., & Reich, J. H. (1991). Competing models of personality disorders.
Journal of Personality Disorders, 5, 60–68. doi:10.1521/pedi.1991.5.1.60 Hersoug, A., Monsen, J., Havik, O., & Høglend, P. (2002). Quality of early working alliance in
psychotherapy: Diagnoses, relationship and intrapsychic variables as predictors. Psychotherapy
and Psychosomatics, 71, 18–27. doi:10.1159/000049340 Holt-Lundstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk:
A meta-analytic review. PloS Medicine, 7, e1000316. doi:10.1371/journal.pmed.1000316
Horneland, M., Sande, D. B., Høbye, K., Knudsen, H., & Lorentzen, S. (2012). Can the clinician-
researcher gap be bridged? Experiences from a randomized clinical trial in analytic/dynamic
group psychotherapy. Group Analysis, 45, 84–98. doi:10.1177/0533316411424370 Horowitz, L. M., Rosenberg, S. E., Baer, B. A., Ureno, G., & Villasenor, V. S. (1988). Inventory of
interpersonal problems: Psychometric properties and clinical applications. Journal of
Consulting and Clinical Psychology, 56, 885–892. doi:10.1037/0022-006X.56.6.885 Jacobson, N. S., & Revenstorf, D. (1988). Statistics for discussing the clinical significance of
psychotherapy techniques: Issues, problems, and new developments. Behavioural
Assessment, 10, 133–145. Retrieved from http://psycnet.apa.org/psycinfo/1989-17676-001
144 Steinar Lorentzen et al.
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to defining
meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology,
59, 12–19. doi:10.1037/0022-006X.59.1.12 Kraemer, H., Wilson, G. T., Fairburn, C., & Agras, W. S. (2002). Mediators and moderators of
treatment effects in randomized clinical trials. Archives of General Psychiatry, 59, 877–883. doi:10.1001/archpsyc.59.10.877
Lambert, M. J., & Bergin, A. E. (1994). The effectiveness of psychotherapy. In S. L. Garfield & A. E.
Bergin (Eds.), The handbook of psychotherapy and behavior change (pp. 143–189). New York, NY: Wiley.
Levenson, J., Wallace, M., Fournier, J., Rucci, P., & Frank, E. (2012). The role of personality pathology
in depression treatment outcome with psychotherapy and pharmacotherapy. Journal of
Consulting and Clinical Psychology, 80, 719–729. doi:10.1037/a0029396 Lipsey, M. W., & Wilson, D. B. (1993). The efficacy of psychological, educational, and behavioral
treatment. Confirmation from meta-analysis. The American Psychologist, 48, 1181–1209. doi:10.1037/0003-066X.48.12.1181
Lorentzen, S. (2014). Group analytic psychotherapy: Working with affective, anxiety and
personality disorders. London, UK/New York, NY: Routledge.
Lorentzen, S., Bøgwald, K., & Høglend, P. (2002). Change during and after long-term analytic group
psychotherapy. International Journal of Group Psychotherapy, 52, 419–429. doi:10.1521/ ijgp.52.3.419.45511
Lorentzen, S., & Høglend, P. (2008). Moderators of the effects of treatment length in long-term
psychodynamic group psychotherapy. Psychotherapy and Psychosomatics, 77, 321–322. doi:10.1159/000147946
Lorentzen, S., & Ruud, T. (2014). Group therapy in public mental health services: Approaches,
patients and group therapists. Journal of Psychiatric and Mental Health Nursing, 21, 219–233. doi:10.1111/jpm.12072
Lorentzen, S., Ruud, T., Fjeldstad, A., & Høglend, P. (2013). Comparison of short- and long-term
dynamic group psychotherapy: Randomised clinical trial. The British Journal of Psychiatry,
203, 280–287. doi:10.1192/bjp.bp.112.113688 McMain, S. F., Guimond, T., Streiner, D. L., Cardish, R. J., & Links, P. S. (2012). Dialectical behavior
therapy compared with general psychiatric management for borderline personality disorder:
Clinical outcomes and functioning over a 2-year follow-up. The American Journal of Psychiatry,
169, 650–661. doi:10.1176/appi.ajp.2012.11091416 Monsen, J. T., Kallerud, J. E., Eilertsen, D. E., & von der Lippe, A. L. (1999). Multisenterprosjekt for
studier av prosess og utfall i psykoterapi [Multi-center project for the study of process and
outcome in psychotherapy]. Oslo, Norway: Institute for Psychology, University of Oslo.
Orlinsky, D. E., Grawe, K., & Parks, B. K. (1994). Process and outcome in psychotherapy: Noch
einmal. In S. L. Garfield & A. E. Bergin (Eds.), Handbook of psychotherapy and behavior change
(pp. 270–376). New York, NY: Wiley. Orlinsky,D.E.,Rønnestad,M.H.,&Willutzki,U.(2004).Fiftyyearsofpsychotherapyprosess-outcome
research: Continuity and change. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook of
psychotherapy and behavior change (pp. 307–389). New York, NY: Wiley. Perry, J. C., Banon, E., & Ianni, F. (1999). Effectiveness of psychotherapy for personality disorders.
The American Journal of Psychiatry, 156, 1312–1321. Retrieved from http://ajp. psychiatryonline.org/journal.aspx?journalid=13
Piper, W. E. (1994). Client variables. In A. Fuhriman & G. M. Burlingame (Eds.), Handbook of
group psychotherapy: An empirical and clinical synthesis (pp. 83–113). New York, NY: Wiley.
Reich, J. H., & Green, A. I. (1991). Effect of personality disorders on outcome of treatment. The
Journal of Nervous and Mental Disease, 179, 74–82. doi:10.1097/00005053-199102000-00003 R€uger, U., & Bell, K. (2004). Historische Entwicklung und aktueller Stand der Richtlinien-
Psychotherapie in Deutschland (The history of psychotherapy in the German system of
Personality disorder moderates therapy outcome 145
compulsory national health insurance). Zeitschrift f€ur Psychosomatische Medizin und Psychotherapie, 50, 127–152. Retrieved from http://www.v-r.de/de
Schulz, K., & Grimes, D. (2002). Blinding in randomised trials: Hiding who got what. Lancet, 359,
696–700. doi:10.1016/S0140-6736(02)07816-9 Sheehan, D., Janavs, J., Baker, R., Harnett-Sheehan, K., Knapp, E., & Sheehan, M. (2002). Mini
international neuropsychiatric interview. Tampa, USA: University of South Florida. Unpublished.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin, 86, 420–428. doi:10.1037/0033-2909.86.2.420 Soeteman, D. R., Verheul, R., & Busschbach, J. J. V. (2008). The burden of disease in personality
disorders: Diagnosis-specific quality of life. Journal of Personality Disorders, 22, 259–268. doi:10.1521/pedi.2008.22.3.259
SPSS (2008). SPSS version 16.0. Chicago, IL: SPSS.
Stiles, W. B., & Shapiro, D. A. (1994). Disabuse of the drug metaphor: Psychotherapy process-
outcome correlations. Journal of Consulting and Clinical Psychology, 62, 942–948. doi:10. 1037/0022-006X.62.5.942
Svartberg, M., Stiles, T. C., & Seltzer, M. H. (2004). Randomized, controlled trial of the effectiveness
of short-term dynamic psychotherapy and cognitive therapy for cluster c personality disorders.
The American Journal of Psychiatry, 161, 810–817. doi:10.1176/appi.ajp.161.5.810 Tschuschke, V., Anbeh, T., & Kiencke, P. (2007). Evaluation of long-term analytic outpatient group
therapies. Group Analysis, 40, 140–159. doi:10.1177/0533316407076127 Verbeke, G., & Molenberg, G. (2000). Linear mixed models for longitudinal data. New York, NY:
Springer.
Verheul, R., & Herbrink, M. (2007). The efficacy of various modalities of psychotherapy for
personality disorders: A systematic review of the evidence and clinical recommendations.
International Review of Psychiatry, 19, 25–38. doi:10.1080/09540260601095399 Vinnars, B., Barber, J. P., Noren, K., Gallop, R., & Weinryb, R. M. (2005). Manualized supportive-
expressive psychotherapy versus nonmanualized community-delivered psychodynamic therapy
for patients with personality disorders: Bridging efficacy and effectiveness. The American Journal
of Psychiatry, 162, 1933–1940. doi:10.1176/appi.ajp.162.10.1933
Received 27 September 2013; revised version received 1 April 2014
146 Steinar Lorentzen et al.
Copyright of British Journal of Clinical Psychology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
