The Benish et al. (2008)
meta-analysis excluded a large number of the comparisons from randomized controlled trials included in the previous meta-analyses. Only 17 comparisons from 15 studies remained. In comparison, Cloitre's (2009)
review lists 44 head-to-head comparisons of face-to-face treatment from 27 studies that were published up to early 2007, the time period reviewed by Benish et al. (2008)
. Benish et al. state that their search of the literature identified 26 comparisons from 22 studies. This raises the question of whether selection procedures in the Benish et al. (2008)
study may have introduced bias. We will examine this question by looking at the way the meta-analysis dealt with non-directive therapies.
Supportive (non-directive, Rogerian, person-centered) therapy is currently widely offered to patients with PTSD in clinical practice. In the British National Health Service, it is the treatment most commonly offered to PTSD patients identified in primary care (e.g., Ehlers, Gene-Cos, & Perrin, 2009
). It is also widely practiced in the United States. Pingitore, Scheffler, Haley, Sentell, and Schwalm (2001)
found that 58% of psychologists practicing in California reported that they provided supportive psychotherapy. There is a good rationale for using supportive therapy to treat PTSD as social support has been shown to be one of the best predictors of recovery in PTSD (Ozer, Best, Lipzey, & Weiss, 2003
). It is, therefore, surprising that most of the trials using such therapies were excluded from the Benish et al. (2008)
meta-analysis. The authors justified the exclusion by arguing that the treatments used in the trials were “not intended to be therapeutic.” This judgment was made even if the trial showed that the treatment was effective (i.e., superior to a no treatment control condition, e.g., Blanchard et al., 2003
In trials of non-directive treatments, two different labels were used to describe the treatment conditions, (1) supportive therapy or supportive counseling and (2) present-centered therapy. All trials using the former label were excluded from the Benish et al. (2008)
meta-analysis. The authors use Foa et al.'s (1991)
study to justify the argument that the supportive therapies used in the research trials were not intended to be therapeutic. In the Foa et al. study (and in a study by Neuner, Schauer, Klaschik, Karunakara, & Elbert, 2004
), therapists in the supportive counseling condition were instructed to steer patients away from talking about their specific traumatic events. We agree that this would not necessarily be representative of supportive therapy as it would be delivered by a practicing clinician, and may therefore underestimate the effect of counseling. However, this restriction did not apply to two other excluded studies (Blanchard et al., 2003; Bryant, Moulds, Guthrie, Dang, & Nixon, 2003
). As shown in , these two studies (but not the Foa et al., 1991, and Neuner et al., 2004
, studies) showed substantial recovery rates with supportive therapy, but nevertheless found that supportive therapy was less effective than TFCBT.
Fig. 1 Comparison of individual non-directive treatments with trauma-focused CBT programs. Intent-to-treat analyses for percent remitted (loss of PTSD diagnosis) with treatment. The study marked with an arrow was selected for Benish et al.'s (2008) meta-analysis. (more ...)
On the other hand, Benish et al. (2008)
judged present-centered therapy (PCT) to be a bona fide
treatment. This treatment aims to control for non-specific therapeutic factors common to active psychotherapies. Therapists help patients identify current life problems and discuss them in a supportive, non-directive mode. PCT as delivered in the trials included an explicit rationale for focusing on the present, psychoeducation about PTSD symptoms, and homework assignments (e.g., Schnurr et al., 2007
). Many of these treatment components were also included in the supportive therapies excluded from the Benish et al. meta-analysis (e.g., Blanchard et al., 2003
). The reasons for the classification difference remain unclear.
Benish et al.'s (2008)
meta-analysis included two studies comparing PCT and TFCBT, a study of individual treatment of survivors of childhood sexual abuse (CSA, McDonagh et al., 2005
; see ), and a study of group
therapy for Vietnam veterans (Schnurr et al., 2003
). Both of these patient populations differ in a number of respects from those included in most other randomized controlled trials of PTSD treatments in that the patients had experienced multiple and prolonged traumas that happened many years ago, are often considered difficult-to-treat, and may require additional interventions (Cloitre, 2009
). In the Schnurr et al. (2003)
Vietnam veterans study, both group treatments led to modest symptom change in the overall intent-to-treat analysis, which included participants who did not receive any treatment. Schnurr et al. (2003)
also presented an analysis of patients who received an adequate dose of group therapy (at least 24 sessions). This analysis suggested that “TFGT (trauma-focused group therapy
) was better than PCGT (present
-centered group therapy
) for treating avoidance and numbing, and possibly, overall PTSD symptoms.” (p. 487). Either way, the fact that this was a group-based treatment makes it difficult to compare with the other studies that were restricted to individual treatment.
In the childhood sexual assault sample (McDonagh et al., 2005
), both individual TFCBT and PCT were superior to the wait list condition at post treatment. In addition, the results pointed to an advantage of CBT at the 3-months follow-up in the completer analysis in that TFCBT participants (82%) were significantly more likely than those receiving PCT (42%) to no longer meet criteria for a PTSD diagnosis.
The pattern of results points to several possible interpretations. One possibility is, as Benish et al. (2008)
suggest, that PCT is as effective as TFCBT. The second possibility is that the lack of differences in the intent-to-treat analyses in the McDonagh et al. (2005) and Schnurr et al. (2003)
studies may be a function of the difficult-to-treat multiple trauma populations studied.
If Benish et al. (2008)
are correct and PCT is equivalent to TFCBT, then one would predict further comparisons of these treatments in other patient populations to show equivalence. If current PTSD treatment guidelines are correct and trauma-focus matters, then one would predict further comparisons to show that TFCBT is superior.
Two recent trials are relevant for deciding between these hypotheses. First, a further trial comparing PCT and TFCBT was omitted from Benish et al.'s (2008)
meta-analysis. Schnurr et al. (2007)
compared PCT with Prolonged Exposure (a form of TFCBT) in female veterans with PTSD. This study found that TFCBT was more effective than PCT (see ). Another trial (Ehlers et al., in preparation
) that was not available at the time of the review compared emotion-focused supportive therapy, which shares many of the active elements of PCT and allowed patients to decide what they wanted to talk about, with Cognitive Therapy for PTSD (a form of TFCBT). Emotion-focused supportive therapy was shown to the effective (superior to wait list). Nevertheless, TFCBT was superior (see ). Thus, both of these studies are at odds with Benish et al.'s (2008)
conclusions and in line with the interpretation that trauma-focus matters and therefore further support the conclusions of recent treatment guidelines that trauma-focused psychological treatments have an advantage over non-directive treatments (Australian Centre for Posttraumatic Mental Health, 2007; National Institute of Clinical Excellence, 2005; Stein et al., 2009
In summary, when one considers all studies comparing individual non-directive therapies with individual TFCBT, it is clear that TFCBT performs better (see ). By excluding nearly all of these studies, Benish et al. (2008)
arrive at the conclusion that there is no difference. Further selectivity is evident in the quotes from published studies reproduced in the paper to support of the equal efficacy argument. illustrates this point.
Table 1 Comparison of results cited in Benish et al. (2008) with authors' conclusions for studies comparing trauma-focused cognitive behavior therapy with other treatments.