Evidence for psychodynamic psychotherapy in specific disorders

disorders: a systematic review

Falk Leichsenring* and Susanne Klein

Clinic of Psychosomatics and Psychotherapy, Justus-Liebig-University Giessen,

Giessen, Germany

(Received 30 October 2013; accepted 3 November 2013)

This article reviews the empirical evidence for psychodynamic therapy for

specific mental disorders in adults. According to the results presented here,

there is evidence from randomized controlled trials (RCTs) that

psychodynamic therapy is efficacious in common mental disorders, including

depressive disorders, anxiety disorders, somatoform disorders, personality

disorders, eating disorders, complicated grief, posttraumatic stress disorder

(PTSD), and substance-related disorders. These results clearly contradict

assertions repeatedly made by representatives of other psychotherapeutic

approaches claiming that psychodynamic psychotherapy is not empirically

supported. However, further research is required, both on outcome and

processes of psychodynamic psychotherapy. There is a need, for example, for

RCTs of psychodynamic psychotherapy of PTSD. Furthermore, research on

long-term psychotherapy for specific mental disorders is required.

To cite this article: Falk Leichsenring & Susanne Klein (2014) Evidence for
psychodynamic psychotherapy in specific mental disorders: a systematic review,
Psychoanalytic Psychotherapy, 28:1, 4-32, DOI: 10.1080/02668734.2013.865428
To link to this article: http://dx.doi.org/10.1080/02668734.2013.865428

In this article, the available evidence for psychodynamic psychotherapy (PDT) in
adults is reviewed. The focus will be on randomized controlled trials (RCTs),
which are regarded as the ‘gold standard’ for demonstrating treatment efficacy.
Previous reviews have been undertaken, for example, by Fonagy, Roth, and
Higgitt (2005), Leichsenring, Klein, and Salzer (in press), Shedler (2010), and
Gerber et al. (2011). Shedler (2010) came to the conclusion that effect sizes of
PDT are as large as those reported for other forms of psychotherapy that are
regarded as ‘empirically supported.’ In addition, he found that effects of PDT
were stable or tended to improve after the end of treatment. In a quality-based
review of RCTs, Gerber et al. (2011) found PDT to be at least as efficacious as
another active treatment in 34 of 39 studies (87%). In comparison with inactive
conditions, PDT was superior in 18 of 24 adequate comparisons (75%).

In another quality-based review of RCTs, Thoma et al. (2012) examined the
methodological quality of RCTs of cognitive-behavioral therapy (CBT) in
q 2014 The Association for Psychoanalytic Psychotherapy in the NHS
*Corresponding author. Email: falk.leichsenring@psycho.med.uni-giessen.de
Psychoanalytic Psychotherapy, 2014
Vol. 28, No. 1, 4–32, http://dx.doi.org/10.1080/02668734.2013.865428
Downloaded by [50.106.28.180] at 09:00 23 March 2014
depression. Contrary to their expectation, the authors found no significant
differences in methodological quality between RCTs of CBT in depression and
RCTs of PDT. Taking the frequently put forward criticism of the methodological
quality of studies of PDT into account (e.g., Bhar & Beck, 2009), the result
reported by Thoma et al. (2012) is of some importance. In another context, we
showed that often double standards were applied when studies of PDT were
criticized by representatives of other approaches (Leichsenring & Rabung, 2011).
Evidence-based medicine and empirically supported treatments
Several proposals have been made to grade the available evidence of both
medical and psychotherapeutic treatments (Canadian Task Force on the Periodic
Health Examination, 1979; Chambless & Hollon, 1998; Clarke & Oxman, 2003;
Cook, Guyatt, Laupacis, Sacket, & Goldberg, 1995; Nathan & Gorman, 2002).

Apart from other differences, all available proposals regard RCTs (efficacy
studies) as the ‘gold standard’ for the demonstration that a treatment is effective.
According to this view, only RCTs can provide level I evidence, which is the
highest level of evidence. RCTs are conducted under controlled experimental
conditions, allowing one to control for variables systematically influencing the
outcome apart from the treatment. The defining feature of an RCT is the random
assignment of subjects to the different conditions of treatment (Shadish, Cook, &
Campbell, 2002). Randomization is regarded as indispensable in order to ensure
that a priori existing differences between subjects are equally distributed. The
goal of randomization is to attribute the observed effects exclusively to the
applied therapy. Thus, randomization is used to ensure the internal validity of a
study (Shadish et al., 2002). Gabbard, Gunderson, and Fonagy (2002) discuss
different types of RCTs that provide different levels of evidence. The most
stringent test of efficacy is achieved by comparison with rival treatments, thus
controlling for specific and unspecific therapeutic factors (Chambless & Hollon,
1998, p. 8). Furthermore, such comparisons provide explicit information
regarding the relative benefits of competing treatments. Treatments that are found
to be superior to rival treatments are more highly valued.

As RCTs are carried out under controlled experimental conditions, their
internal validity is usually high. However, for this very reason, their external
validity may be limited, in that their results may not be fully representative of
clinical practice. In contrast to RCTs, naturalistic studies (observational or
effectiveness studies) are conducted under the conditions of clinical practice.
Thus, their results are usually more representative for clinical practice with
regard to patients, therapists, and treatments (external validity). RCTs and
observational studies address different questions of research, i.e., efficacy under
controlled experimental conditions versus effectiveness under the conditions of
clinical practice (Leichsenring, 2004). For this reason, RCTs are not ‘bad’ and
observational studies are not ‘good’ or vice versa. Their relationship is
complementary rather than one of rival (Leichsenring, 2004).


Methods
Definition of PDT
PDT operates on an interpretive-supportive continuum (Gunderson & Gabbard,
1999; Wallerstein, 1989). Interpretive interventions enhance the patient’s insight
about repetitive conflicts sustaining his or her problems (Gabbard, 2004; Luborsky,
1984). Supportive interventions aim to strengthen abilities (‘ego-functions’) that
are temporarily not accessible to a patient due to acute stress (e.g., traumatic
events) or that have not been sufficiently developed (e.g., impulse control in
borderline personality disorder; BPD). Thus, supportive interventions maintain or
build ego functions (Wallerstein, 1989). Supportive interventions include, for
example, fostering a therapeutic alliance, setting goals, or strengthening ego
functions such as reality testing or impulse control (Luborsky, 1984). The use of
more supportive or more interpretive (insight-enhancing) interventions depends on
the patient’s needs. The more severely disturbed a patient is, or the more acute his
or her problem is, the more supportive and less interpretive interventions are
required and vice versa (Luborsky, 1984; Wallerstein, 1989). Borderline patients,
as well as healthy subjects, in an acute crisis or after a traumatic event may need
more supportive interventions (e.g., stabilization, providing a safe and supportive
environment). Thus, a broad spectrum of psychiatric problems and disorders can be
treated with PDT, ranging from milder adjustment disorders or stress reactions to
severe personality disorders such as BPD or psychotic conditions.

Inclusion and exclusion criteria
The following inclusion and exclusion criteria were applied: (1) PDT according
to the definition above was applied, (2) RCT, (3) reliable and valid measures for
diagnosis and outcome, (4) use of treatment manuals, and (5) study of specific
mental disorders. Studies examining the combination of psychodynamic therapy
and medication were not included, however, concomitant medication in both
treatment arms was allowed.
We collected studies of PDT that were published between 1970 and
September 2013 by use of a computerized search of MEDLINE, PsycINFO, and
Current Contents. The following search terms were used: (psychodynamic or
dynamic or psychoanalytic*) and (therapy or psychotherapy or treatment) and
(study or studies or trial*) and (outcome or result* or effect* or change*) and
(psych* or mental*) and (RCT* or control* or compare*). Manual searches in
articles and textbooks were performed. In addition, we communicated with
authors and experts in the field.
Efficacy studies of PDT in specific mental disorders
A total of 47 RCTs providing evidence for the efficacy of PDT in specific mental
disorders were identified and included in this review. These studies are presented
in Table 1.

Models of PDT
In the studies identified, different forms of PDT were applied (Table 1). The
models developed by Luborsky (1984), Shapiro and Firth (1985), and Malan
(1976) were used most frequently.

Evidence for the efficacy of PDT in specific mental disorders
The studies of PDT included in this review will be presented for different
mental disorders. However, from a psychodynamic perspective, the results of a
therapy for a specific psychiatric disorder (e.g., depression, agoraphobia) are
influenced by the underlying psychodynamic features (e.g., conflicts, defenses,
personality organization), which may vary considerably within one category
of psychiatric disorder (Kernberg, 1996). These psychodynamic factors may
affect treatment outcome and may have a greater impact on outcome than
the phenomenological DSM categories (Piper, McCallum, Joyce, Rosie, &
Ogrodniczuk, 2001).

Depressive disorders
At present, several RCTs are available that provide evidence for the efficacy of
PDT compared to CBT in major depressive disorder (Barkham et al., 1996;
Driessen et al., 2013; Gallagher-Thompson & Steffen, 1994; Shapiro et al., 1994;
Thompson, Gallagher, & Breckenridge, 1987). It is of note that due to the large
sample size the RCT by Driessen et al. (2013) was sufficiently powered for an
equivalence trial. Different models of PDT were applied (Table 1). Thase (2013)
concluded from this RCT: ‘On the basis of these findings, there is no reason to
believe that psychodynamic psychotherapy is a less effective treatment of major
depressive disorder than CBT.’

In another RCT by Salminen et al. (2008), PDT was found to be equally
efficacious as fluoxetine in reducing symptoms of depression and improving
functional ability. However, with sample sizes of N1 ¼ 26 and N2 ¼ 25, statistical
power may have not been sufficient to detect possible differences between
treatments. In a small RCT, Maina, Forner, and Bogetto (2005) examined the
efficacy of PDT and brief supportive therapy in the treatment of minor depressive
disorders (dysthymic disorder, depressive disorder not otherwise specified, or
adjustment disorder with depressed mood). Both treatments were superior to a
waiting-list condition at the end of treatment. At six-month follow-up, PDT was
superior to brief supportive therapy. In a recent study by Barber, Barrett, Gallop,
Rynn, and Rickels (2012), PDT and pharmacotherapy were equally effective in
the treatment of depression. However, neither PDT nor pharmacotherapy was
superior to placebo.

An earlier meta-analysis (Leichsenring, 2001) found PDT and CBT to be
equally effective with regard to depressive symptoms, general psychiatric
symptoms, and social functioning. These results are consistent with the findings
12 F. Leichsenring and S. Klein
Downloaded by [50.106.28.180] at 09:00 23 March 2014
of more recent meta-analyses by Barth et al. (2013) and Driessen et al. (2010;
Abbass & Driessen, 2010). Barth et al. (2013) did not find significant differences
in outcome between different forms of psychotherapy of depression. Driessen
et al. (2010) found PDT significantly superior to control conditions. If group
therapy was included, PDT was less efficacious compared to other treatments at
the end of therapy. If only individual therapy was included, there were no
significant differences between PDT and other treatments (Abbass & Driessen,
2010). In three-month and nine-month follow-ups, no significant differences
between treatments were found.

Meanwhile, internet-guided self-help is also available for PDT. In an RCT,
Johansson et al. (2012) found internet-guided self-help based on PDT
significantly more efficacious than a structured support intervention (psychoeducation
and scheduled weekly contacts online) in patients with major depressive
disorder. Treatment effects were maintained at 10-month followup.
Psychodynamically oriented self-help was based on the concept by
Silverberg (2005). Silverberg’s internet-guided self-help based on PDT is a
promising approach, especially for patients who do not receive psychotherapy.
Further studies should be carried out.
In summary, several RCTs provide evidence for the efficacy of PDT in
depressive disorders.

Pathological grief
In two RCTs by McCallum and Piper (1990) and Piper et al. (2001), the treatment
of prolonged or complicated grief by short-term psychodynamic group therapy
was studied. In the first study, short-term psychodynamic group therapy was
significantly superior to a waiting list (McCallum & Piper, 1990). In the second
study, a significant interaction was found. With regard to grief symptoms,
patients with high quality of object relations improved more in interpretive
therapy, and patients with low quality of object relations improved more in
supportive therapy. For general symptoms, clinical significance favored
interpretive therapy over supportive therapy (Piper et al., 2001).

Anxiety disorders
For anxiety disorders, several RCTs are presently available (Table 1). With
regard to panic disorder (with or without agoraphobia), Milrod et al. (2007)
showed in an RCT that PDT was more successful than applied relaxation. For
social phobia, three RCTs of psychodynamic therapy exist. In the first study,
short-term psychodynamic group treatment for generalized social phobia was
superior to a credible placebo control (Knijnik, Kapczinski, Chachamovich,
Margis, & Eizirik, 2004).

In a study by Bo¨gels, Wijts, and Sallerts (2003), PDT proved to be as effective
as CBT in the treatment of (generalized) social phobia. However, with sample
sizes of N ¼ 22 and N ¼ 24, statistical power may have not been sufficient to
detect possible differences between treatments.
In a large-scale multicenter RCT, the efficacy of PDT and cognitive therapy
(CT) in the treatment of social phobia was studied (Leichsenring et al., 2013a). In
an outpatient setting, 495 patients with a primary diagnosis of social phobia were
randomly assigned to CT, PDT, or the waiting list. Treatments were carried out
according to manuals and treatment fidelity was carefully controlled for. Both
treatments were significantly superior to the waiting list. Thus, this trial provides
evidence that PDT is effective in the treatment of social phobia according to the
criteria proposed by Chambless and Hollon (1998). There were no differences
between PDT and CT with regard to response rates for social phobia (52% vs.
60%) and reduction of depression. There were significant differences between CT
and PDT in favor of CT, however, with regard to remission rates (36% vs. 26%),
self-reported symptoms of social phobia, and reduction of interpersonal
problems. Differences in terms of between-group effect sizes, however, were
small and below the priori set threshold for clinical significance (Leichsenring,
Salzer, & Leibing, in press; Leichsenring et al., 2013a). Taking these results
referring to clinically significant differences into account, recommending CBT
over PDT in social anxiety disorders is not warranted. As Kraemer (2011,
p. 1350) puts it: ‘Only if the ES [effect size] is greater than some value d*
[threshold of clinical significance] is a strong clinical recommendation of one
treatment over the other warranted.’ For the comparison of PDT with CBT, this
was not the case. Furthermore, in the follow-up study 6, 12, and 24 months after
end of therapy, neither statistically significant nor clinically significant
differences were found between CT and PDT in any outcome measure
(Leichsenring et al., 2013b). In general, the differentiation between statistical and
clinical significance has not yet been sufficiently taken into account in
psychotherapy research. From small, but statistically significant differences, the
conclusion is drawn that one treatment is superior to another (Leichsenring et al.,
in press).

In a randomized controlled feasibility study of generalized anxiety disorder,
PDT was equally effective as a supportive therapy with regard to continuous
measures of anxiety, but significantly superior on symptomatic remission rates
(Crits-Christoph, Connolly Gibbons, Narducci, Schamberger, & Gallop, 2005).
However, the sample sizes of that study were relatively small (N ¼ 15 vs.
N ¼ 16), and the study was not sufficiently powered to detect more possible
differences between treatments. In another RCT of generalized anxiety disorder,
PDT was compared to CBT (Leichsenring et al., 2009). PDT and CBT were
equally effective with regard to the primary outcome measure. However, in some
secondary outcome measures, CBT was found to be superior, both at the end of
therapy and at the six-month follow-up. Other differences may exist that were not
detected due to the limited sample size and power (CBT: N ¼ 29; PDT: N ¼ 28).
In the one-year follow-up, results proved to be stable (Salzer, Winkelbach,
Leweke, Leibing, & Leichsenring, 2011). 

The core element in the applied method of CBT consisted of a modification of
worrying. This specific difference between the treatments may explain the
superiority of CBT in the Penn State Worry Questionnaire (Meyer, Miller,
Metzger, & Borkovec, 1990) and, in part, also in the State-Trait Anxiety
Inventory (trait measure) (Spielberger, Gorsuch, & Lushene, 1970) – the latter
also contains several items related to worrying. The results of that study may
suggest that the outcome of STPP in generalized anxiety disorder may be further
optimized by employing a stronger focus on the process of worrying. In PDT,
worrying can be conceptualized as a mechanism of defense that protects the
subject from fantasies or feelings that are even more threatening than the contents
of his or her worries (Crits-Christoph, Wolf-Palacio, Ficher, & Rudick, 1995).
According to the available RCTs, PDT is efficacious in anxiety disorders. If
differences between PDT and CBT were found, they showed up in secondary
outcome measures or corresponded to small differences in effect size. This is
consistent with a recent meta-analysis by Baardseth et al. (2013) who did not find
significant differences in favor of CBT compared to bona fide treatments.

For CBT, a recent historical review showed that the efficacy of treatments for
anxiety disorders has not increased but rather decreased from the 1980s to the
present (O¨ st, 2008). Furthermore, a substantial proportion of patients do not
sufficiently benefit from the treatments and the proportion of nonresponders does
not appear to have decreased over time (O¨ st, 2008). For these reasons, there is a
need to further improve the treatment of anxiety disorders (Schmidt, 2012). This
is true not just for CBT, but also for PDT as well (Leichsenring, Klein, Salzer,
2014). In one of the most promising approaches to address this problem,
psychotherapy research is moving from single-disorder-focused manualized
approaches toward ‘transdiagnostic’ and modular treatments (e.g., Barlow, Allen,
& Choate, 2004; McHugh, Murray, & Barlow, 2009). The rationale for
transdiagnostic treatments focuses on similarities among disorders, particularly
in a similar class of diagnoses (e.g., anxiety disorders), including high rates of
comorbidity and improvements in comorbid conditions when treating a principal
disorder (Barlow et al., 2004; McHugh et al., 2009). For these reasons,
researchers in the field of CBT have developed transdiagnostic treatment
protocols (e.g., Barlow et al., 2004; McHugh et al., 2009; Norton & Phillip,
2008). It is an advantage that PDT is traditionally less tailored to single mental
disorders, but focuses on core underlying processes of mental disorders. A recent
review has shown that the empirically supported methods of PDT for specific
anxiety disorders have core treatment components in common (Leichsenring &
Salzer, in press). These components have been distilled and integrated into an
evidence-based Unified Psychodynamic Protocol for ANXiety disorders (UPPAnx;
Leichsenring & Salzer, in press)
.
Integrating treatment elements of empirically supported methods of PDT for
specific anxiety disorders, the manualized UPP-Anx has the potential to: (1) be
more effective than single-disorder psychotherapy, (2) be more effective than
routine PDT, (3) improve comorbid symptoms, (4) enhance patients’ quality of
life, (5) facilitate translation of research into clinical practice of mental health
professionals, (6) facilitate training for practitioners and dissemination of the
approach relative to training in several distinct single-disorder treatments, (7) be
more cost efficient (e.g., by additionally improving comorbid symptoms), and (8)
have an impact on both the health-care system and public health. As a next step,
we are planning to evaluate the UPP-Anx in a RCT.

Mixed samples of depressive and anxiety disorders
Knekt et al. (2008a, 2008b) compared STPP, long-term psychodynamic
psychotherapy (LTPP), and solution-focused therapy (SFT) in patients with
depressive or anxiety disorders. STPP was more effective than LTPP during the
first year. During the second year of follow-up, no significant differences were
found between long-term and short-term treatments. In the three-year follow-up,
LTPP was more effective; no significant differences were found between the
short-term treatments. With regard to specific mental disorders, it is of note that
after three years significantly more patients recovered from anxiety disorders in
LTPP (90%) compared to STPP (67%) and SFT (65%). For depressive disorders,
no such differences occurred. In an RCT by Bressi, Porcellana, Marinaccio,
Nocito, and Magri (2010), PDT was superior to Treatment as Usual (TAU) in a
sample of patients with depressive or anxiety disorders.

Posttraumatic stress disorder
In an RCT by Brom, Kleber, and Defares (1989), the effects of PDT, behavioral
therapy, and hypnotherapy in patients with posttraumatic stress disorder (PTSD)
were studied. All of the treatments proved to be equally effective. The results
reported by Brom et al. (1989) are consistent with that of a more recent metaanalysis
by Benish, Imel, and Wampold (2008), which found no significant
differences between bona fide treatments of PTSD. In a response to the metaanalysis
by Benish et al. (2008), Ehlers et al. (2010) critically reviewed the study
by Brom et al. (1989). A comprehensive discussion with a convincing reply to the
critique by Ehlers et al. (2010) was given by Wampold et al. (2010). In the present
context, we shall only address the critique put forward by Ehlers et al. (2010)
against the study by Brom et al. (1989). Ehlers et al. (2010) reviewed the study by
Brom et al. (1989) in the following way (p. 273, italics by the authors): ‘In this
study, neither hypnotherapy nor psychodynamic therapy was consistently more
effective than the waiting-list control condition across the analyses used . . .’ In
addition, Brom et al. (1989) pointed out that ‘Patients in psychodynamic therapy
showed slower overall change than those in the other two treatment conditions,
and did not improve in intrusive symptoms significantly . . .’
Results are different for different outcome measures. For the avoidance scale
and the total score of the Impact of Event Scale (Horowitz, Wilner, & Alvarez,
1979), PDT was significantly superior to the waiting-list condition, both after
therapy and at follow-up (Brom et al., 1989, p. 610, Table 1). While effect sizes
for PDT were somewhat smaller at posttreatment (avoidance: 0.66, total: 1.10),
PDT achieved the largest effect sizes at follow-up (avoidance: 0.92, total: 1.56)
as compared to CBT (avoidance: 0.73, total: 1.30) and hypnotherapy (avoidance:
0.88, total: 1.54).

For the intrusion scale of the Impact of Event Scale, the primary outcome
measure, it is true that PDT was not superior to waiting list both at posttest and
at three-month follow-up. Intrusion is one of the core symptoms of PTSD. Pre–
post differences of PDT, however, were significant and the pre–post and prefollow-
up effect sizes were large (0.95 and 1.55, respectively). In contrast, the
pre–post effect size for the waiting list was small (0.34). For the CBT condition
(trauma desensitization), the pre–post and pre-follow-up effect sizes were 1.66
and 1.43, respectively. Thus, at follow-up, PDT achieved a larger effect size
than CBT. While the effect size of CBT tended to decrease at follow-up, it
tended to increase for PDT; as will be shown below, this is true for the
avoidance scale and the total score of the Impact of Event Scale. For this reason,
it is strange that the difference between PDT and the control condition was
reported by Brom et al. (1989) to be not significant at follow-up. For intrusion,
PDT achieved the lowest score of all conditions at follow-up. These results,
however, were not reported by Ehlers et al. (2010). The figure presented by
Ehlers et al. (2010, p. 273, Figure 2) only included the pre–post effect sizes, but
not the pre-follow-up effect sizes, for which PDT achieved larger effect sizes,
as shown above. In a critical review, results of all analyses should be presented,
not only the results that support one’s own perspective. Furthermore, for
general symptoms, Brom et al. (1989) wrote that PDT ‘seems to withstand the
comparison [with waiting list] best’ (p. 610). Thus, after all, it seems to take (a
little bit, i.e., three months!) longer for PDT to achieve its effects, but these
effects are at least as large as those of CBT.
Further studies of PDT in PTSD are required. At present, only one RCT of
PDT in PTSD is presently available.

Somatoform disorders
At present, five RCTs of PDT in somatoform disorders that fulfill the inclusion
criteria are available (Table 1). In the RCT by Guthrie, Creed, Dawson, and
Tomenson (1991), patients with irritable bowel syndrome, who had not
responded to standard medical treatment over the previous six months, were
treated with PDT in addition to standard medical treatment. This treatment was
compared to standard medical treatment alone. According to the results, PDT was
effective in two-thirds of the patients. In another RCT, PDT was significantly
more effective than routine care, and as effective as medication (paroxetine) in
the treatment of severe irritable bowel syndrome (Creed et al., 2003). During the
follow-up period, however, PDT, but not paroxetine, was associated with a
significant reduction in health-care costs compared with TAU. In an RCT by
Hamilton et al. (2000), PDT was compared to supportive therapy in the treatment
of patients with chronic intractable functional dyspepsia, who had failed to
respond to conventional pharmacological treatments. At the end of treatment,
PDT was significantly superior to the control condition. The effects were stable in
the 12-month follow-up.

An RCT by Faramarzi et al. (2013) corroborated these results with PDT
combined with medical treatment being superior to medical treatment alone, with
regard to gastrointestinal symptoms, defense mechanisms, and alexithymia, both
at the end of therapy and at the 1- and 12-month follow-up. Monsen and Monsen
(2000) compared PDT of 33 sessions with a control condition (no treatment or
TAU) in the treatment of patients with chronic pain. PDT was significantly
superior to the control group on measures of pain, psychiatric symptoms,
interpersonal problems, and affect consciousness. The results remained stable or
even improved in the 12-month follow-up. In a recent study, Sattel et al. (2012)
compared PDT with enhanced medical care in patients with multi-somatoform
disorders. At follow-up, PDT was superior to enhanced medical care with regard
to improvements in patients’ physical quality of life.

Abbass, Kisely, and Kroenke (2009) carried out a review and meta-analysis
on the effects of PDT in somatic disorders. They included both RCTs and
controlled before and after studies. Meta-analysis was possible for 14 studies. It
revealed significant effects on physical symptoms, psychiatric symptoms, and
social adjustment, which were maintained in long-term follow-up. Thus,
specific forms of PDT can be recommended for the treatment of somatoform
disorders.

Bulimia nervosa
For the treatment of bulimia nervosa, three RCTs of PDT are available (Table 1).
Significant and stable improvements in bulimia nervosa after PDT were
demonstrated in the RCTs by Fairburn, Kirk, O’Connor, and Cooper (1986),
Fairburn et al. (1995), and Garner et al. (1993). In the primary disorder-specific
measures (bulimic episodes, self-induced vomiting), PDT was as effective as
CBT (Fairburn et al., 1986, 1995; Garner et al., 1993). Again, however, the
studies were not sufficiently powered to detect possible differences (see Table 1
for sample sizes). Apart from this, CBT was superior to PDT in some specific
measures of psychopathology (Fairburn et al., 1986). However, in a follow-up
(Fairburn et al., 1995) of the Fairburn et al. (1986) study using a longer follow-up
period, both forms of therapy proved to be equally effective and were partly
superior to a behavioral form of therapy. Accordingly, for a valid evaluation of
the efficacy of PDT in bulimia nervosa, longer-term follow-up studies are
necessary. In another RCT, PDT was significantly superior to both a nutritional
counseling group and CT (Bachar, Latzer, Kreitler, & Berry, 1999). This was true
of patients with bulimia nervosa and a mixed sample of patients with bulimia
nervosa or anorexia nervosa.

Anorexia nervosa
For the treatment of anorexia nervosa, however, evidence-based treatments are
barely available (Fairburn, 2005). This applies to both PDT and CBT. In an RCT
by Gowers, Norton, Halek, and Crisp (1994), PDT combined with four sessions
of nutritional advice yielded significant improvements in patients with anorexia
nervosa (Table 1). Weight and body mass index (BMI) changes were
significantly more improved than in a control condition (TAU). Dare, Eisler,
Russell, Treasure, and Dodge (2001) compared PDT with a mean duration of 24.9
sessions to cognitive-analytic therapy, family therapy, and routine treatment in
the treatment of anorexia nervosa (Table 1). PDT yielded significant
symptomatic improvements and PDT and family therapy were significantly
superior to the routine treatment with regard to weight gain. However, the
improvements were modest – several patients were undernourished at the followup.
A recent RCT compared manual-guided psychodynamic therapy, enhanced
CBT, and optimized TAU in the treatment of anorexia nervosa (Zipfel et al.,
2013). After 10 months of treatment, significant improvements were found in all
treatments, with differences in the primary outcome measure (BMI). At the 12-
months follow-up, however, psychodynamic therapy was significantly superior to
optimized TAU, whereas enhanced CBT was not (Zipfel et al., 2013). Recovery
rates were 35% versus 19% versus 13% for psych