by Michael D. Anestis, M.S.
Of the topics we cover on PBB, one of my favorites is comparative treatment trials. In these studies, multiple forms of therapy are compared to one another in the treatment of a diagnosis or set of diagnoses, thereby allowing us to get a sense of whether one is better than the other and, as such, should be prioritized as an option for clients in need of help for a particular condition. Today, I would like to discuss a new study that fits within this category. Written by Michel Dugas of Concordia University and Hopital du Sacre-Coeur de Montreal and colleagues (in press) and soon to be published in Behavior Therapy, the study looked at the utility of a specific form of cognitive behavioral therapy (CBT) and applied relaxation (AR) in the treatment of generalized anxiety disorder (GAD).
In the past, we have discussed a number of treatment trials for individuals with GAD. In one, CBT was shown to outperform psychodynamic therapy on measures of worry, depression, and trait anxiety (click here to read our coverage of that study). In another, whereas CBT outperformed AR in the treatment of panic disorder, the two treatments were equally efficacious in the treatment of GAD (click here for our coverage of that study). The study I am going to write about today examined a newer form of CBT - and remember, CBT is a class of treatments, not one specific form of therapy that is identical across or even within certain diagnoses - developed by the authors. Their version of CBT is based upon the theory that individuals with GAD are characterized by an inability to tolerate uncertainty. In other words, when faced with ambiguous information or unknown variables, individuals with GAD become fixated on potential outcomes, resulting in pathological worry. For instance, if an individual with GAD does not know whether or not they will receive a promotion, they may start worrying excessively about things related to that topic, then spiral off into worries completely irrelevant to the topic at hand, in large part because they can not tolerate their lack of certainty. Their model also emphasizes the tendency for individuals with GAD to believe that worrying is a useful activity, to develop a negative problem orientation, and to avoid processing difficult cognitive information. Prior research on the authors' particular version of GAD-specific CBT has provided impressive results:
- CBT outperformed wait-list control group on all outcomes and both short- and long-term outcomes were equal to the best outcomes for other forms of therapy (Ladouceur et al., 2000)
- Group format CBT outperformed wait-list control group on all outcome measures and individuals who received group CBT decreased their levels of worry from post-treatment to 24-month follow-up, meaning that they continued to improve on the primary outcome measure in the two years following the end of therapy (Dugas et al., 2003)
- CBT outperformed nondirective psychotherapy on diagnostic remission and symptom reduction and was more effective at helping long-term benzodiazepine users discontinue medication use (Gosselin, Ladouceur, Morin, Dugas, & Baillargeon, 2006)
These results are impressive, however, the authors noted that none of those studies involved the comparison of their version of CBT to another active treatment with empirical support. In other words, it's nice to know that their version of CBT is better than nothing, but it would be more useful to know that it is at least as good as other treatments for which we already have empirical support indicating that they are useful in the treatment of GAD.
To address this, the authors conducted a randomized controlled study (RCT), in which 65 participants were randomized to receive either CBT or AR or were placed in a wait-list no treatment condition. All treatments involved weekly 1-hour sessions for 12 weeks. After the 12 weeks, participants in the waitlist condition were randomized to one of the two treatment conditions, so nobody went without treatment. Overall, the participants had an average age of 38.5 and had been experiencing GAD for 13.9 years. All participants included in the study had a primary diagnosis of GAD as determined first by a screening interview and next by a full structured diagnostic interview and 58.5% of the sample also met criteria for at least one other mental illness. This is important as many critics of RCTs and research in general argue that the clients included in such studies are not like those in "real world" practice due to a lack of comorbidity. All of the participants in this study were adults and nobody was included who demonstrated suicidal intent, current substance abuse, or current or past schizophrenia, bipolar disorder, or organic mental disorder. Additionally, all participants had to demonstrate significantly greater severity in their primary diagnosis (GAD) than in any comorbid diagnosis (see the original study for details on this point). The sample thus obviously was not perfect, but it was not nearly as restrictive as many other trials (additionally, click here to read an earlier article we wrote about data indicating that comorbidity does not have any negative impact on treatment outcomes for GAD).
Before conducting the study itself, the authors wanted to address some issues that might otherwise make it difficult to interpret the results. First of all, they wanted to account for common factors by having clients fill out a series of questionnaires addressing their impression of such variables. Secondly, they wanted to account for allegiance effects. Because the authors themselves developed this form of CBT, it is possible that, given the opportunity, they might unwittingly act in a way that would unfairly favor that treatment approach and result in inaccurate findings. To avoid that, they hired a practicing clinical psychologist who had been trained primarily in psychodynamic therapy and had no extensive training in either of the treatments used in this study. She received training in both approaches by experts prior to the onset of the study and was given weekly supervision in both approaches throughout the course of the study and the degree to which she implemented the treatments as directed by the protocols was assessed by an independent observer (results indicated strong treatment fidelity in both conditions). Also, treatment outcomes were assessed by individuals who were unaware of which treatment condition participants were in. So, by hiring a therapist who was not trained in one particular form of treatment and had nothing to gain by favoring one versus the other, they reduced the likelihood that therapy would be conducted in a manner that favored one treatment over the other. By independently assessing treatment fidelity, they ensured that the therapist, who again had little training in either approach, actually implemented each treatment as it was designed. By having independent raters assess outcomes without knowledge of treatment conditions, they ensured that individuals in one condition were not said to have improved more or less in a manner that would artificially favor one treatment over the other. This is an extremely impressive degree of effort on the part of the authors to remove bias from the study and to therefore provide accurate results that can generalize to other populations outside context of a controlled study.
A total of two participants dropped out of the CBT condition and 5 dropped out of AR. For those individuals, their scores at pre-treatment were used as their scores at post-treatment as well, which is a conservative way of making sure that the results of the study do not only reflect the positive outcomes of people who manage to stay in therapy all the way to the end.
When comparing each of the two treatments to the wait-list, the authors found that:
- CBT outperformed the wait-list on GAD symptom severity, worry, somatic anxiety, and overall clinical improvement, but did not outperform wait-list on depression or trait anxiety.
- AR outperformed the wait-list on GAD symptom severity, but not on any other outcomes.
Importantly, the authors noted that the individuals in the wait-list improved on their own prior to treatment much more so than is typically seen in GAD treatment trials. As such, the differences between treatment and no-treatment may have been artificially diminished, which would explain why AR did not outperform the wait-list as it has in several other studies.
When comparing CBT directly to AR after individuals on the wait-list had been randomized to treatment and received therapy, they found:
- The two treatments were equal at post-treatment on all outcomes except one, with CBT outperforming AR on overall clinical improvement.
When examining long-term outcomes, which included assessments at 6, 12, and 24 months post-treatment, the authors found:
- The two treatments resulted in equivalent overall results, but individuals who received CBT demonstrated continued improvement on trait anxiety, worry, and overall clinical improvement whereas individuals who received AR did not continue to improve as time went by post-treatment.
Ultimately, the authors provided us with clear evidence that their CBT protocol is better than a wait-list control condition and equivocal evidence that it outperforms AR. The upside to this finding is that the results support the notion that there is more than one efficacious treatment for GAD. The downside is that the unusually large improvement by individuals on the wait-list may have obscured the results and rendered them somewhat difficult to interpret. The authors should be commended for going to such great lengths to ensure that the results were not influenced by biases and for not excluding individuals with any comorbid diagnosis. In the end, the authors concluded that they might need to adjust their CBT protocol in order to enhance its ability to outperform other treatments, but they were pleased with the overall positive results of the study. Keep in mind that, of all the anxiety disorders, GAD is the one for which we have the least impressive treatment outcomes, which adds an additional layer of complexity when running and interpreting a trial like this one.
What I like most about this study is that it directly compared two treatments that already had strong empirical support on a number of outcome variables. So, rather than indirectly comparing how the two treatments did versus another treatment (or lack of treatment) they added an analysis that directly compared them to one another. Such studies are not perfect, but I believe they provide a clearer evidence base than do meta-analyses, which combine studies and make comparisons that involve less impressive studies.
If you would like to learn more about generalized anxiety disorder or its treatment, we recommend the following items, each of which are available through our online store for scientifically-based psychological resources:
- Generalized Anxiety Disorder: Advances in Research and Practice
by Richard Heimberg, Cynthia Turk, and Doug Mennin
- Cognitive-Behavioral Treatment for Generalized Anxiety Disorder: From Science to Practice by Michel Dugas and Melisa Robichaud