Previous meta-analyses in the field of depressive disorders have yielded mixed conclusions about which the most efficacious treatment is.  Within the process of any therapy, there are universal features such as; the therapeutic relationship, rationale for treatment, a patient and therapist belief in treatment, a confidential and respectful space for the patient and so on.  However, individual to each psychotherapy there will also be specific factors such as for example; the use of behavioural experiments in CBT or the focus on the patient/therapist relationship in psychodynamic therapy.

Braun, Gregor & Traun (2012) set out to complete four tasks:

1. Update evidence on bona fide psychotherapies for depression
2. Clarify and investigate the status of non-CBT bona fide therapies in comparison to CBT
3. Compare two meta-analytical approaches (conventional random-effects model and a network meta-analysis)
4. To identify moderators that may explain relative efficacy of different psychotherapies

Bona fide therapies were identified by three criteria:

1. That therapists were trained to provide the treatment and held adequate qualifications
2. That therapists met face to face with the person to deliver an individualised treatment (i.e. not following a standard protocol)
3. That the treatment contained “psychologically valid components”

Methods

Using a systematic search strategy the reviewers identified 53 papers which were appropriate for inclusion.  Studies included were published between 1977 and 2012, including 3,965 patients of which 70.6% were female.

Criteria for inclusion were that the studies had to compare efficacy of psychotherapy treatment for depressive disorder with another psychotherapy treatment in a randomised trial, and that at least two treatments that were compared were bona fide.

Study population (comorbidity, age, gender, clinical diagnosis etc.), treatment (number and length of sessions, therapy format, therapist training etc.) and study characteristics (randomisation, study quality, researcher allegiance etc.) were considered as possible moderators and coded by reviewers (where double coded there was a 93% agreement). Study quality (assessed using a risk bias assessment tool) and researcher allegiance were only coded once.

Independent effect sizes (converted to Hedges g in order to adjust bias due to small sample size; 0.20 is considered small, 0.50 medium, and 0.80 large) were calculated for the five outcome measures:

1. Patient rated Beck Depression Inventory (BDI)
2. Clinician rated Hamilton Depression Rating Scale (HDRS)
3. Combined outcome measure based on mean of patient rated BDI and clinician rated HDRS
4. Ratio of odds (OR) of remission
5. OR of dropout

Results

Meta-analysis I included only therapies where there were 5 or more comparisons available and found:

• There is no strong indication of superiority of CBT, Behavioural Activation (BA), Psychodynamic Therapy (DYN), and Interpersonal Therapy (IPT) (range of g= -0.09 to 0.19) however Supportive Therapies (SUP) according to Rogerian theory were shown to be less efficacious than other therapies (g=0.26)
• There was no evidence of therapy (CBT, BA, DYN or IPT) superiority in relation to remission rates (OR range: 0.87 to 1.02; OR>1 indicates higher likelihood of remission), however Supportive Therapies (SUP) were less efficacious (OR=0.65, p<0.01)

• Moderator analysis:
  • Participants over the age of 60 appeared to benefit more from BA than other treatments and BA was more efficacious when delivered in an individual rather than group/couples setting
  • According to clinician ratings, CBT was more efficacious than other therapies for females
  • CBT was more efficacious than other treatments when therapy lasted 90 minutes or longer, whereas BA was more efficacious when therapy sessions lasted less than 90 minutes. Longer sessions were more likely to be associated with a group format delivery

Meta-analysis II allowed for analysis of previous grouped together SUP therapies and using CBT as a comparison treatment:

• None of the comparisons between CBT and other therapies reached nominal significance, however reviewers commented on trends which indicated some efficacy in “other therapies” including cognitive behavioural analysis system of psychotherapy (CBASP) and problem solving therapy (PST), albeit based on a small number of comparisons
• Comparison of results from meta-analysis I and II indicate different results across patient, clinician and combined outcome measures on CBT vs. SUP, BA, DYN and IPT.  Results show larger effect sizes in the network meta-analysis (II).

• 

  According to the moderator analysis, behavioural activation was more effective than other treatments for older people

Conclusions

The authors concluded:

Our results suggest that the dodo bird verdict seems to be the right answer for the wrong question.  Even though it seemed mostly corroborated at the aggregate level, there appear to exist a number of differential effects in efficacy between bona fide treatments of depression at a finer level.

The results of the meta-analyses showed small effect sizes for the differences between psychotherapies, effectively suggesting that (of the therapies included in this review) there is not one therapy which stands out as yielding better outcomes.  However, results do throw up some interesting questions about the moderating effects of; participant age, therapy format, patient gender, nature of disorder and comorbidity.

The limitations of this review impede any firm conclusions being drawn:

• There were not enough studies to conduct separate meta-analyses for each type of treatment.  Overall there were three times as many CBT comparisons (k, number of comparisons=41) than any other therapy BA (k=16), DYN (k=11), IPT (k=10), SUP (k=17)
• As well as a small number of study comparisons there was variability within the sample which is relatively unclear (e.g. patient diagnosis and comorbid presentations) and some potentially relevant details are not reported (e.g. depressive episode or duration of illness)
• The use of network meta-analyses may not have been appropriate given the diversity of the study samples as it is more likely to make assumptions about the data when looking at both direct and indirect comparisons

In essence the most comprehensive conclusion to be drawn is that more research into psychotherapies for depression is needed.  This would allow for more therapy specific meta-analyses which might yield results that have more clinical significance.  This is no mean feat however, after all, this review highlights the challenge of being able to define and measure the many nuances of therapy and their relative contribution to each therapy session.

Link

Braun SR, Gregor B, Tran US (2013) Comparing Bona Fide Psychotherapies of Depression in Adults with Two Meta-Analytical Approaches. PLoS ONE 8(6): e68135. doi:10.1371/journal.pone.0068135