**2. What is characteristic in PTSD treatments?**

Psychotherapy outcome research aims at identifying treatment components that are critical for symptom improvement. The RCT design has been adopted from medical research and became standard in psychotherapy outcome research. This design relies on the assumption that the overall treatment effect is composed of first, the true effect of the treatment under investigation and second, effects that are due to the context of being in treatment.

Whereas the first type of components has typically been described as 'specific' or 'active' components in psychotherapy outcome research, for the second type of components a number of synonymously used terms occur in the literature, although they may have slightly different connotations [17, 18]. Such terms include 'common,' 'general' or 'non-specific' factors or 'psychological placebos.' For the present review, we will follow the terminology proposed by Grünbaum ([19], p. 159). Grünbaum's definition captures the outlined dichotomy with reference to the presence or absence of a psychological (e.g. etiological) theory, which defines the content of a complex treatment, such as psychotherapy. Accordingly, treatment components will be considered *characteristic* if there is a theoretical model that describes how the respective component will contribute to symptom improvement. In contrast, treatment components will be considered *incidental* if there is no such theory-based link to symptom improvement. Thus, the components that are considered active or specific would be considered *characteristic*, because typically they are deduced from psychological theories, which describe how they will improve the symptoms of a particular disorder. In contrast to the concept of specificity, which has been related to uniqueness of treatment components [18], components may be considered *characteristic* even though they may not be a unique component of a particular treatment package. All other factors that may contribute to a treatment effect but which have not (yet) been specified within a psychological theory would be considered *incidental*, no matter whether they are common to all treatments, shared by some or not at all related to the treatment itself, but, for instance, rather to the patient, therapist or to the conduct of the study.

see [3]). Until recently, clinical guidelines and systematic reviews concluded that patients with PTSD require psychotherapeutic treatments that specifically target the trauma experience [4]. However, recent meta-analyses showed that focusing on the trauma experience may not be

Randomized controlled trials (RCTs) have typically been conducted in order to identify those components of complex treatment packages that critically impact symptom improvement (i.e. in placebo-controlled studies and comparative or dismantling studies, see [8]). Recently, however, the general validity of RCTs has been criticized in medical as well as in psychotherapy research, by showing that extra-therapeutic factors (such as blinding of outcome assessors or the sample size) may considerably affect the outcome in RCTs [9–12]. Accordingly, metaanalyses, which attempted to explain variation between effect estimates from individual studies—the so-called between-study heterogeneity—identified a number of moderators of

This paper summarizes meta-analytic findings, which show that in PTSD outcome research extra-therapeutic factors affect the outcome. These findings relate to two questions in the current debate in PTSD outcome research: first, 'Is there evidence that some PTSD treatments consistently outperform others?' and second, 'Is a trauma focus generally necessary for successful PTSD treatment?'. We will briefly describe the research designs that have been used in order to address the abovementioned questions. Then, we will describe common flaws in meta-analyses and we will use examples from PTSD research in order to show how flaws in meta-analyses may lead to invalid conclusions. And finally, we will summarize how the consideration of relevant moderators may alter the conclusions that may be drawn from meta-

Psychotherapy outcome research aims at identifying treatment components that are critical for symptom improvement. The RCT design has been adopted from medical research and became standard in psychotherapy outcome research. This design relies on the assumption that the overall treatment effect is composed of first, the true effect of the treatment under

Whereas the first type of components has typically been described as 'specific' or 'active' components in psychotherapy outcome research, for the second type of components a number of synonymously used terms occur in the literature, although they may have slightly different connotations [17, 18]. Such terms include 'common,' 'general' or 'non-specific' factors or 'psychological placebos.' For the present review, we will follow the terminology proposed by Grünbaum ([19], p. 159). Grünbaum's definition captures the outlined dichotomy with reference to the presence or absence of a psychological (e.g. etiological) theory, which defines the content of a complex treatment, such as psychotherapy. Accordingly, treatment components will be considered *characteristic* if there is a theoretical model that describes how the respective component will contribute to symptom improvement. In contrast, treatment

investigation and second, effects that are due to the context of being in treatment.

analyses of RCTs with respect to the two highlighted research questions.

**2. What is characteristic in PTSD treatments?**

generally necessary for successful PTSD treatment [5–7].

230 A Multidimensional Approach to Post-Traumatic Stress Disorder - from Theory to Practice

treatment effects in PTSD RCTs [7, 13–16].

When looking at PTSD research, a number of etiological models of PTSD led to the definition of a number of *characteristic* components and accordingly to the development of several rival treatment packages (see [3, 5, 20, 21]). However, the classification of PTSD treatments according to the underlying etiological model has been a challenge in previous meta-analyses [14]. An inconsistent terminology and the use of treatment labels that are not clearly defined and thus not exclusive (see [22]) lead to considerable variation in the classification of PTSD treatments across individual meta-analyses. In order to reduce some complexity and despite the differences in the foci of the underlying etiological models (e.g. focusing on cognitions vs. focusing on behavioral aspects), several treatment packages have been summarized under the umbrella term of trauma-focused treatments [23]. But again, the definition of the term remained largely unclear and led to inconsistencies with respect to which treatments were to be considered trauma-focused [22]. In contrast, treatments that clearly do not address the trauma experience or even proscribe talking about the trauma have consistently been used as psychological placebo control conditions.

According to the Grünbaum definition, for the present review, we will consider treatments that provided some theory-driven link between treatment components and symptom improvement as *characteristi*c (this includes treatments that have previously been summarized as trauma-focused), whereas the clearly non-trauma-focused interventions would be regarded as relying on *incidental* treatment components. However, we will not be able to resolve the inconsistencies, which appear between different meta-analyses and which are due to the imprecise terminology.
