by Michael D. Anestis, M.S.
An interesting interaction in the comment section on the recent "bad science" article got me thinking about today's topic: dismantling studies. When we design treatments, our underlying belief is that each aspect of treatment contributes to the improvement of the client. In other words, we do not design treatments that we believe include useless activities. That being said, like any belief, this one is subject to empirical investigation and one of the ways that we do this is through dismantling studies.
Let's say, for instance, that it was me rather than Aaron Beck who designed cognitive behavioral therapy (CBT) for depression. Let's also say that, in my version, I included an added component at the end of therapy during which the client ate a ham sandwich - a wonderfully tasty ham sandwich. Let's further say that my theory behind the ham sandwich component was that ham is inherently a positive experience due to its extreme tastiness and that my belief was that having such a wonderful experience at the end of treatment was pivotal in helping the client accept that they had healed and to move forward into their newly happy life with confidence. I suspect that nobody reading this believes my ham theory; however, regardless of the general consensus of the theory's validity, we actually do not have to simply wonder or take somebody's word for it. This is an empirical question that can be tested. We simply need to recruit a sample and randomly assign individuals to either receive the Anestis ham sandwich CBT approach or Beck's hamless CBT and see if there are any differences between the groups. If the ham sandwich was important, the ham group would exhibit greater improvements. If not, there would be no difference - or alternatively, the other group might demonstrate that the ham addition was harmful or at least an obstacle in treatment by exhibiting greater improvement.
Fortunately, several less ridiculous hypotheses have been tested and today I would like to discuss one of them. In a study recently published in the Journal of Consulting and Clinical Psychology, Patricia Resick and colleagues (2008) dismantled cognitive processing therapy (CPT). CPT is one of two well-established empirically supported treatments (EST's) for post-traumatic stress disorder (PTSD), with the other being prolonged exposure (PE). CPT involves two basic components: cognitive therapy aimed at challenging distorted cognitions and altering the meaning of the traumatic event and written accounts (WAs) in which the client writes detailed accounts of the traumatic event and repeatedly reads the description both at home and in session in order to habituate to the anxiety provoked by reminders of the trauma. The question that Resick and her colleagues (2008) wanted to answer was whether the full protocol of CPT was the most effective approach or whether individuals components of treatment would offer equally promising results.
Before describing the study itself, it is important to note that, while dismantling studies provide a lot of value, they are imperfect. When a treatment is designed, the individual components only make up a portion of the treatment. As such, when you compare the full treatment to treatment consisting only of individual components, that means that you are going to adjust the amount of time spent on any particular part of treatment. In other words, if treatment is designed to last 12 sessions, those sessions are generally spent on a variety of skills. If only one component is used, however, the dosage of that component will increase substantially. As a result, when we compare a full treatment to treatment consisting only of particular components, we are not truly making an equal comparison. This imperfection, however, does not detract from the value provided by this type of study.
So, back to the study at hand. The authors' plan was to compare full CPT to cognitive therapy only (CPT-C) and written accounts only (WAs) in order to see if any particular component was particularly useful and if the full protocol offered incremental value over the components individually. All of the participants in the study were women who met criteria for PTSD as determined through a structured diagnostic interview. Participants were excluded if they were illiterate (WA's would not work) or currently psychotic, suicidal, or dependent upon drugs or alcohol. Individuals struggling with substance abuse were permitted to participate if they agreed to abstain throughout the course of treatment. Participants also had to be at least 3 months removed from the traumatic event, as this is the period during which natural improvements independent of therapeutic interventions are most likely to occur and the authors wanted to ensure that their results were not better accounted for by the simple passage of time. The intent-to-treat (ITT) sample included 150 individuals.
The authors were very careful to ensure that the therapists were faithful to the treatment protocol, taping every session, reviewing the sessions, and meeting for weekly group supervision to discuss cases. Participants were randomly assigned to one of the three conditions. For a full description of the therapeutic approaches, I recommend that you read the actual article, but here is a brief description of each condition:
***********
Standard CPT
Session 1: Education about PTSD, overview of treatment, rationale for treatment, assignment to write a description of the meaning of the traumatic event (impact statement)
Session 2: Read and discuss impact statement, learn about relationship between thoughts, behaviors, and emotions
Session 3: Assigned to write detailed description of traumatic event. Clients are encouraged not to avoid emotional responses and to read the description to themselves daily
Session 4: Client reads statement to the therapist and therapist uses cognitive therapy approaches to question distorted cognitions. Client again is assigned to write a description of the event for homework.
Sessions 5-6: Same as session 4, but with greater emphasis on teaching the client to challenge distorted thoughts
Sessions 7-12: Clients use more advanced worksheets to practice challenging thoughts and understanding the relationship between thoughts, emotions, and behaviors. In session 11, the client is asked to again write a description of the event, this time reflecting new beliefs, and the new version is read and discussed during the final session.
Cognitive therapy only (CPT-C)
The protocol for CPT-C was identical to standard CPT, minus the use of written accounts.
Written accounts only (WA)
Whereas in standard CPT, the clients write their WAs at home, in this condition, the clients write the descriptions in session in order to ensure that they take sufficient time and are exposed to thoughts of the trauma for a substantial period of time. Clients are instructed to frequently note their subjective units of distress (SUDs), a measure of how much general distress he or she feels at a given moment - through the writing process in order to chart shifts in negative affect. The client reads the WA to the therapist, but no cognitive therapy approaches are used in response. Instead, the therapist is only allowed to make nondirective supportive comments. Again, the clients are told to rewrite and read the WA to themselves throughout the week as homework.
***********
And the results...
So, what did they find? Actually, the results were a bit surprising. First of all, all three groups showed substantial improvement not only in PTSD symptoms, but also in depression, anxiety, anger, guilt, shame, and cognitive distortions. Additionally, these benefits were maintained at a 6-month follow-up assessment after treatment had ended. The success of all three groups was not particularly surprising, but the success of each group relative to one another was. Full CPT, combining cognitive therapy and WAs, did not result in greater improvements than did either component on its own. In other words, cognitive therapy alone and WAs alone were just as effective as full CPT. On a measure of PTSD symptoms, cognitive therapy outperformed WAs; however, both groups produced impressive reductions in PTSD symptoms.
These findings are important, as they highlight the point that, while CPT is an effective treatment for PTSD, clients might benefit just as much from a protocol that only utilizes WA's or cognitive therapy. Given that this is only one study and that the authors had to adjust the structure of WA's from what is typically done in CPT, we need to be conservative in our interpretations of the data and careful not to assume too much. That being said, Resick and colleagues (2008) did a great job of directly assessing the question of whether or not a therapist needs to administer the entire CPT manual as outlined in order to attain the desired effect in treatment. This early evidence suggests that the protocol can be adjusted to only reflect one component without losing the benefits of the entire protocol, but more research is needed in order to more fully address the question.
Importantly though, be careful not to look at these results an interpret them to mean that therapists can deviate from a protocol at random or simply as they see fit and produce equivalent results. No therapist did this in this study, so there is absolutely no support for that assumption. Deviation from the protocol in this study involved a pre-determined alteration with very specific steps designed prior to the onset of therapy. As such, the idea of being "eclectic" and simply integrating what seems to work in an idiosyncratic manner is not an approach that is supported by this study.
**Edit** I want to send a quick thank you to Dr.Resick. In a thoughtful email, she alerted me to a point in this paper that was not properly emphasized. In most studies on PTSD, the sample sizes and statistical methods used prevent researchers from being able to detect anything other than large effect sizes. In other words, unless the difference between treatments is enormous, the difference is difficult to detect statistically and, as a result, some studies have failed to find expected differences between approaches. Quite simply, it is difficult and expensive to recruit a sufficiently large sample and many researchers do not have the background in statistics to use the most efficient analytical procedures. In this study, however, the sample size and particularly the data analytic procedure allowed for the detection of a smaller, but vital effect: the superior PTSD finding for CPT-C relative to WA. Why is this so important? Because it indicates that extinction (the removal of the fear response to cues related to the traumatic event) is not necessarily the end all be all of PTSD treatment and that cognitive processes may, in fact, be a more important target in treatment. It will be very interesting to see how this plays out in future studies looking at outcomes across treatment approaches.
If you would like to learn more about PTSD or cognitive processing therapy, we recommend the following resources, all of which are available through our online store:
Cognitive Processing Therapy for Rape Victims: A Treatment Manual
by Patricia Resick and Monica Schnicke
Prolonged Exposure Therapy for PTSD: Emotional Processing of Traumatic Experiences Therapist Guide
by Edna Foa, Elizabeth Hembree, and Barbara Olaslov Rothbaum
Prolonged Exposure Therapy for Adolescents with PTSD Emotional Processing of Traumatic Experiences, Therapist Guide
by Edna Foa, Kelly Chrestman, and Eva Gilboa-Schnechtman
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University
Subscribe in a reader
