Johathan Shedler

Page 4 of 8 Previous page Next Page

The Efficacy of Psychodynamic Psychotherapy

Jonathan Shedler
University of Colorado Denver School of Medicine

February–March 2010 American Psychologist
© 2010 American Psychological Association 0003-066X/10/$12.00
Vol. 65, No. 2, 98–109 DOI: 10.1037/a0018378

How Effective Is Psychodynamic Therapy?

A recent and especiallymethodologicallyrigorous meta-analysis of psychodynamic therapy, published by the Cochrane Library,5 included 23 randomized controlled trials of 1,431 patients (Abbass, Hancock, Henderson, & Kisely, 2006). The studies compared patients with a range of common mental disorders6 who received short-term (< 40 hours) psychodynamic therapy with controls (wait list, minimal treatment, or “treatment as usual”) and yielded an overall effect size of 0.97 for general symptom improvement. The effect size increased to 1.51 when the patients were assessed at long-term follow-up (>9monthsposttreatment). In addition tochange in general symptoms, the meta-analysis reported an effect size of 0.81 for change in somatic symptoms, which increased to 2.21 at long-term follow-up; an effect size of 1.08 for change in anxiety ratings, which increased to 1.35 at follow-up; and an effect size of 0.59 for change in depressive symptoms, which increased to 0.98 at follow-up.7 The consistent trend toward larger effect sizes at follow-up suggests that psychodynamic therapy sets in motion psychological processes that lead to ongoing change, even after therapy has ended.

A meta-analysis published in Archives of General Psychiatry included 17 high-quality randomized controlled trials of short-term (average of 21 sessions) psychodynamic therapy and reported an effect size of 1.17 for psychodynamic therapy compared with controls (Leichsenring, Rabung, & Leibing, 2004). The pretreatment to posttreatment effect size was 1.39, which increased to 1.57 at long-term follow-up, which occurred an average of 13 months posttreatment. Translating these effect sizes into percentage terms, the authors noted that patients treated with psychodynamic therapy were “better off with regard to their target problems than 92% of the patients before therapy” (Leichsenring et al., 2004, p. 1213).

A newly released meta-analysis examined the efficacy of short-term psychodynamic therapy for somatic disorders (Abbass, Kisely, & Kroenke, 2009). It included 23 studies involving 1,870 patients who suffered from a wide range of somatic conditions (e.g., dermatological, neurological, cardiovascular, respiratory, gastrointestinal, musculoskeletal, genitourinary, immunological). The study reported effect sizes of 0.69 for improvement in general psychiatric symptoms and 0.59 for improvement in somatic symptoms. Among studies that reported data on health care utilization, 77.8% reported reductions in health care utilization that were due to psychodynamic therapy—a finding with potentially enormous implications for health care reform.

A meta-analysis reported in the American Journal of Psychiatry examined the efficacy of both psychodynamic psychotherapy (14 studies) and CBT (11 studies) for personality disorders (Leichsenring & Leibing, 2003). The metaanalysis reported pretreatment to posttreatment effect sizes using the longest term follow-up available. For psychodynamic therapy (mean length of treatment was 37 weeks), the mean follow-up period was 1.5 years and the pretreatment to posttreatment effect size was 1.46. For CBT (mean length of treatment was 16 weeks), the mean follow-up period was 13 weeks and the effect size was 1.0. The authors concluded that both treatments demonstrated effectiveness. A more recent review of short-term (average of 30.7 sessions) psychodynamic therapy for personality disorders included data from seven randomized controlled trials (Messer & Abbass, in press). The study assessed outcome at the longest follow-up period available (an average of 18.9 months posttreatment) and reported effect sizes of 0.91 for general symptom improvement (N = 7 studies) and 0.97 for improvement in interpersonal functioning (N = 4 studies).

Two recent studies examined the efficacy of long-term psychodynamic treatment. A meta-analysis reported in the Journal of the American Medical Association (Leichsenring & Rabung, 2008) compared long-term psychodynamic therapy (> 1 year or 50 sessions) with shorter term therapies for the treatment of complex mental disorders (defined as multiple or chronic mental disorders, or personality disorders) and yielded an effect size of 1.8 for overall outcome.8 The pretreatment to posttreatment effect size was 1.03 for overall outcome, which increased to 1.25 at long-term follow-up (p < .01), an average of 23 months posttreatment. Effect sizes increased from treatment completion to follow-up for all five outcome domains assessed in the study (overall effectiveness, target problems, psychiatric symptoms, personality functioning, and social functioning). A second meta-analysis, reported in the Harvard Review of Psychiatry (de Maat, de Jonghe, Schoevers, & Dekker, 2009), examined the effectiveness of long-term psychodynamic therapy (average of 150 sessions) for adult outpatients with a range of diagnoses. For patients with mixed/moderate pathology, the pretreatment to posttreatment effect was 0.78 for general symptom improvement, which increased to 0.94 at long-term follow-up, an average of 3.2 years posttreatment. For patients with severe personality pathology, the pretreatment to posttreatment effect was 0.94, which increased to 1.02 at long-term follow-up, an average of 5.2 years posttreatment.

These meta-analyses represent the most recent and methodologically rigorous evaluations of psychodynamic therapy. Especially noteworthy is the recurring finding that the benefits of psychodynamic therapy not only endure but increase with time, a finding that has now emerged from at least five independent meta-analyses (Abbass et al., 2006; Anderson & Lambert, 1995; de Maat et al., 2009; Leichsenring & Rabung, 2008; Leichsenring et al., 2004). In contrast, the benefits of other (nonpsychodynamic) empirically supported therapies tend to decay over time for the most common disorders (e.g., depression, generalized anxiety; de Maat, Dekker, Schoevers, & de Jonghe, 2006; Gloaguen, Cottraux, Cucharet, & Blackburn, 1998; Hollon et al., 2005; Westen, Novotny, & Thompson-Brenner, 2004).9

Table 1

Table 1 summarizes the meta-analytic findings described above and adds additional findings to provide further points of reference. Except as noted, effect sizes listed in the table are based on comparisons of treatment and control groups and reflect response at the completion of treatment (not long-term follow-up).
Studies supporting the efficacy of psychodynamic therapy span a range of conditions and populations. Randomized controlled trials support the efficacy of psychodynamic therapy for depression, anxiety, panic, somatoform disorders, eating disorders, substance-related disorders, and personality disorders (Leichsenring, 2005; Milrod et al., 2007).

Findings concerning personality disorders are particularly intriguing. A recent study of patients with borderline personality disorder (Clarkin, Levy, Lenzenweger, & Kern-berg, 2007) not only demonstrated treatment benefits that equaled or exceeded those of another evidence-based treatment, dialectical behavior therapy (Linehan, 1993), but also showed changes in underlying psychological mechanisms (intrapsychic processes) believed to mediate symptom change in borderline patients (specifically, changes in reflective function and attachment organization; Levy et al., 2006). These intrapsychic changes occurred in patients who received psychodynamic therapy but not in patients who received dialectical behavior therapy.

Such intrapsychic changes may account for long-term treatment benefits. A newly released study showed enduring benefits of psychodynamic therapy five years after treatment completion (and eight years after treatment initiation). At five-year follow-up, 87% of patients who received “treatment as usual” continued to meet diagnostic criteria for borderline personality disorder, compared with 13% of patients who received psychodynamic therapy (Bateman & Fonagy, 2008). No other treatment for personality pathology has shown such enduring benefits.
These last findings must be tempered with the caveat that they rest on two studies and therefore cannot carry as much evidential weight as findings replicated in multiple studies conducted by independent research teams. More generally, it must be acknowledged that there are far more empirical outcome studies of other treatments, notably CBT, than of psychodynamic treatments. The discrepancy in sheer numbers of studies is traceable, in part, to the indifference to empirical research of earlier generations of psychoanalysts, a failing that continues to haunt the field and that contemporary investigators labor to address.

A second caveat is that many psychodynamic outcome studies have included patients with a range of symptoms and conditions rather than focusing on specific diagnostic categories (e.g., those defined by diagnostic criteria specified in the Diagnostic and Statistical Manual of Mental Disorders [4th edition, DSM-IV; American Psychiatric Association, 1994]). The extent to which this is a limitation is open to debate. A concern often raised about psychotherapy efficacy studies is that they use highly selected and unrepresentative patient samples and, consequently, that their findings do not generalize to real-world clinical practice (e.g., Westen et al., 2004). Nor is there universal agreement that DSM–IV diagnostic categories define discrete or homogeneous patient groups (given that psychiatric comorbidity is the norm and that diagnosable complaints are often embedded in personality syndromes; Blatt & Zuroff, 2005; Westen, Gabbard, & Blagov, 2006). Be that as it may, an increasing number of studies of psychodynamic treatments do focus on specific diagnoses (e.g., Bateman & Fonagy, 2008; Clarkin et al., 2007; Cuijpers, van Straten, Andersson, & van Oppen, 2008; Leichsenring, 2001, 2005; Milrod et al., 2007).


5 More widely known in medicine than in psychology, the Cochrane Library was created to promote evidence based practice and is considered a leader in methodological rigor for meta-analysis.

6 These included nonpsychotic symptom and behavior disorders commonly seen in primary care and psychiatric services, for example, nonbipolar depressive disorders, anxiety disorders, and somatoform disorders, often mixed with interpersonal or personality disorders (Abbass et al., 2006).

7 The meta-analysis computed effect sizes in a variety of ways. The findings reported here are based on the single method that seemed most conceptually and statistically meaningful (in this case, a random effects model, with a single outlier excluded). See the original source for more fine-grained analyses (Abbass et al., 2006).

8 The atypical method used to compute this effect size may provide an inflated estimate of efficacy, and the effect size may not be comparable to other effect sizes reported in this review (for discussion, see Thombs, Bassel, & Jewett, 2009).

9 The exceptions to this pattern are specific anxiety conditions such as panic disorder and simple phobia, for which short-term, manualized treatments do appear to have lasting benefits (Westen et al., 2004).